Email alert
2018 Vol. 40, No. 5
Display Method:
2018, 40(5): 1017-1023.
doi: 10.11999/JEIT170772
Abstract:
A secret key generation scheme is proposed in a Single Carrier Frequency Domain Equalization system (SC-FDE). The scheme uses the Channel Frequency Response (CFR) to generate the secret keys, which is termed as CFR-Key. The principle of the CFR-Key scheme is introduced and formulate the secret key capacity is derived based on mutual information theory. The Signal-Noise Ratio (SNR) is proved that is the unique factor and affects the quantization levels. The optimum quantization levels is designed to achieve the best key generation rate when SNR is given. Simulation results show that comparing with the secret key generation scheme based on the Channel Impulse Response (CIR-Key), the proposed scheme can significantly improve the secret key capacity.
A secret key generation scheme is proposed in a Single Carrier Frequency Domain Equalization system (SC-FDE). The scheme uses the Channel Frequency Response (CFR) to generate the secret keys, which is termed as CFR-Key. The principle of the CFR-Key scheme is introduced and formulate the secret key capacity is derived based on mutual information theory. The Signal-Noise Ratio (SNR) is proved that is the unique factor and affects the quantization levels. The optimum quantization levels is designed to achieve the best key generation rate when SNR is given. Simulation results show that comparing with the secret key generation scheme based on the Channel Impulse Response (CIR-Key), the proposed scheme can significantly improve the secret key capacity.
2018, 40(5): 1024-1030.
doi: 10.11999/JEIT170671
Abstract:
Considering the energy-constrained cooperative cognitive radio networks, the maximization problem of Energy Efficiency (EE) with Quality of Service (QoS) constraint of primary user is investigated. Secondary user receives the signals of primary user using Simultaneous Wireless Information and Power Transfer (SWIPT) and serves as a relay with decode-and-forward protocol. Based on fractional programming and introducing auxiliary variables, the original optimization issue is converted to a convex issue and solved, and an iterative algorithm for resource allocation is proposed. Simulation results demonstrate that the proposed algorithm converges to the optimal solution quickly. Compared with the energy-cooperation strategy, the proposed strategy achieves a higher EE and protects the QoS of primary user better.
Considering the energy-constrained cooperative cognitive radio networks, the maximization problem of Energy Efficiency (EE) with Quality of Service (QoS) constraint of primary user is investigated. Secondary user receives the signals of primary user using Simultaneous Wireless Information and Power Transfer (SWIPT) and serves as a relay with decode-and-forward protocol. Based on fractional programming and introducing auxiliary variables, the original optimization issue is converted to a convex issue and solved, and an iterative algorithm for resource allocation is proposed. Simulation results demonstrate that the proposed algorithm converges to the optimal solution quickly. Compared with the energy-cooperation strategy, the proposed strategy achieves a higher EE and protects the QoS of primary user better.
2018, 40(5): 1031-1036.
doi: 10.11999/JEIT170751
Abstract:
In order to improve the Simultaneous Wireless Information and Power Transfer (SWIPT) efficiency, the Base Station (BS) employs the Three-Dimensional (3D) directional antennas and exploits the vertical domain by dynamically adjusting the antenna tilt. So the efficiency of energy and information transfer can be increased. The single-cell Multiple-Input Multiple-Output (MIMO) SWIPT system is studied where the BS applies Zero-Forcing (ZF) precoding and users use the Power Splitting (PS) technique. The optimization problem is formulated to minimize the transmit power subject to SNR and harvested power targets. And the antenna tilt, PS ratios and the allocated power of each user are jointly optimized. The optimal PS ratios and the optimal power allocation are given in closed-form expressions. Simulations show that the proposal outperforms conventional MIMO systems with an adjustable tilt and the Two-Dimensional (2D) scheme without considering the vertical domain.
In order to improve the Simultaneous Wireless Information and Power Transfer (SWIPT) efficiency, the Base Station (BS) employs the Three-Dimensional (3D) directional antennas and exploits the vertical domain by dynamically adjusting the antenna tilt. So the efficiency of energy and information transfer can be increased. The single-cell Multiple-Input Multiple-Output (MIMO) SWIPT system is studied where the BS applies Zero-Forcing (ZF) precoding and users use the Power Splitting (PS) technique. The optimization problem is formulated to minimize the transmit power subject to SNR and harvested power targets. And the antenna tilt, PS ratios and the allocated power of each user are jointly optimized. The optimal PS ratios and the optimal power allocation are given in closed-form expressions. Simulations show that the proposal outperforms conventional MIMO systems with an adjustable tilt and the Two-Dimensional (2D) scheme without considering the vertical domain.
2018, 40(5): 1037-1043.
doi: 10.11999/JEIT170731
Abstract:
In order to overcome the problem of ignoring the sensing information between the two fixed thresholds for double-thresholds energy detection algorithm, the sequential cooperative sensing algorithm based on dynamic adaptive double-threshold energy detection is proposed. Since the soft decision between the two dynamic adaptive double thresholds is used for optimizing the probability of detection, it takes sequential cooperative sensing algorithm to model the adaptive double-threshold regions. Furthermore, the proposed algorithm can adaptively adjust the thresholds values and the number of collaborative users for each decision region to optimize the probability of detection and receiver operating characteristics curve. Simulation results show that the proposed algorithm has better performance for the detection capability, compared with other double thresholds sensing algorithms.
In order to overcome the problem of ignoring the sensing information between the two fixed thresholds for double-thresholds energy detection algorithm, the sequential cooperative sensing algorithm based on dynamic adaptive double-threshold energy detection is proposed. Since the soft decision between the two dynamic adaptive double thresholds is used for optimizing the probability of detection, it takes sequential cooperative sensing algorithm to model the adaptive double-threshold regions. Furthermore, the proposed algorithm can adaptively adjust the thresholds values and the number of collaborative users for each decision region to optimize the probability of detection and receiver operating characteristics curve. Simulation results show that the proposed algorithm has better performance for the detection capability, compared with other double thresholds sensing algorithms.
2018, 40(5): 1044-1049.
doi: 10.11999/JEIT170647
Abstract:
Sparse Code Multiple Access (SCMA) is a non-orthogonal multiple access technology based on multi-dimensional codebook, which can effectively address challenges in 5G such as massive connectivity, high spectral efficiency and millisecond delay. For the problem that threshold-based Message Passing Algorithm (MPA) has a high Bit Error Rate (BER) when the threshold is low, an improved SCMA multiuser detection scheme is proposed in this paper. Based on the threshold MPA scheme, the proposed scheme adds a judgment on the necessary conditions of user node stability. The users who not only satisfy the threshold criterion but also pass the judgment on the necessary conditions of user node stability can be decoded in advance. This improves the reliability of codeword which is judged in advance and reduces the loss of posterior soft information caused by the detection mechanism similar to the hard decision. Compared with the threshold-based MPA scheme, the proposed scheme allows the messages to be iterated more fully at low thresholds, which makes SCMA users achieve better BER performance at low thresholds. The simulation results show that better BER performance is achieved with the proposed scheme than that with the threshold-based MPA scheme for SCMA users.
Sparse Code Multiple Access (SCMA) is a non-orthogonal multiple access technology based on multi-dimensional codebook, which can effectively address challenges in 5G such as massive connectivity, high spectral efficiency and millisecond delay. For the problem that threshold-based Message Passing Algorithm (MPA) has a high Bit Error Rate (BER) when the threshold is low, an improved SCMA multiuser detection scheme is proposed in this paper. Based on the threshold MPA scheme, the proposed scheme adds a judgment on the necessary conditions of user node stability. The users who not only satisfy the threshold criterion but also pass the judgment on the necessary conditions of user node stability can be decoded in advance. This improves the reliability of codeword which is judged in advance and reduces the loss of posterior soft information caused by the detection mechanism similar to the hard decision. Compared with the threshold-based MPA scheme, the proposed scheme allows the messages to be iterated more fully at low thresholds, which makes SCMA users achieve better BER performance at low thresholds. The simulation results show that better BER performance is achieved with the proposed scheme than that with the threshold-based MPA scheme for SCMA users.
2018, 40(5): 1050-1058.
doi: 10.11999/JEIT170781
Abstract:
This papers propose a novel integrated Wi-Fi and Micro Electronic Mechanical Systems (MEMS) indoor mobility map construction and localization approach. First of all, a method is proposed for constructing mobility map based on trajectory main path by applying the Pedestrian Dead Reckoning (PDR), Minimum Description Length (MDL), and Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithms to the processing process of crowdsourcing trajectories. Then a pixel template matching technique is innovatively presented to obtain the absolute position of the map. Finally, the robust Extended Kalman Filter (EKF) algorithm is utilized to estimate the optimal target position. Which means the Simultaneous Localization And Mapping (SLAM) are completed. The experimental results show that the method of proposed clustering can accurately distinguish the motion regions. Also, the precision positioning can be realized with less labor and time through matching the absolute position of the motion map in the real environment successfully.
This papers propose a novel integrated Wi-Fi and Micro Electronic Mechanical Systems (MEMS) indoor mobility map construction and localization approach. First of all, a method is proposed for constructing mobility map based on trajectory main path by applying the Pedestrian Dead Reckoning (PDR), Minimum Description Length (MDL), and Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithms to the processing process of crowdsourcing trajectories. Then a pixel template matching technique is innovatively presented to obtain the absolute position of the map. Finally, the robust Extended Kalman Filter (EKF) algorithm is utilized to estimate the optimal target position. Which means the Simultaneous Localization And Mapping (SLAM) are completed. The experimental results show that the method of proposed clustering can accurately distinguish the motion regions. Also, the precision positioning can be realized with less labor and time through matching the absolute position of the motion map in the real environment successfully.
2018, 40(5): 1059-1065.
doi: 10.11999/JEIT170655
Abstract:
In the Time Division Long Term Evolution (TD-LTE) indoor distributed network, the signal differences among different positions are insignificant, thus accurate localization can not be achieved by reference point calibration. To solve this compelling problem, this paper proposes a correlation sequencing based localization algorithm. Firstly, the signal information among adjacent positions is utilized to construct Reference Signal Receiving Power (RSRP) motion sequence database. Next, correlation sequencing algorithm is conducted to obtain relation between real-time RSRP sequence and the ones in constructed database, which results in a set of candidate sequence. After that, the correlation coefficient and mean Euclidean distance between candidate sequences and the online one are calculated. Finally, the optimal candidate sequence is selected by a voting strategy to estimate targets position. Experimental results show that the proposed localization algorithm can effectively improve the localization accuracy within indoor distributed antenna system.
In the Time Division Long Term Evolution (TD-LTE) indoor distributed network, the signal differences among different positions are insignificant, thus accurate localization can not be achieved by reference point calibration. To solve this compelling problem, this paper proposes a correlation sequencing based localization algorithm. Firstly, the signal information among adjacent positions is utilized to construct Reference Signal Receiving Power (RSRP) motion sequence database. Next, correlation sequencing algorithm is conducted to obtain relation between real-time RSRP sequence and the ones in constructed database, which results in a set of candidate sequence. After that, the correlation coefficient and mean Euclidean distance between candidate sequences and the online one are calculated. Finally, the optimal candidate sequence is selected by a voting strategy to estimate targets position. Experimental results show that the proposed localization algorithm can effectively improve the localization accuracy within indoor distributed antenna system.
2018, 40(5): 1066-1071.
doi: 10.11999/JEIT170672
Abstract:
This paper proposes a new method of authentication based on piling up tag signal. Using the advantage of legality sides sharing a private key and having coherent channel state information in short time, the sender produces a tag signal from spread frequency code and measures channel state information, then overlaps the tag signal on the communicational signal, the receiver can detect the correct tag signal and demodulate the communicational signal. This method avoids using complex cryptographic algorithm, reduces the amount of calculation, and prevents passive eavesdropping and active attack effectively. Simulation results show that this method has high use value.
This paper proposes a new method of authentication based on piling up tag signal. Using the advantage of legality sides sharing a private key and having coherent channel state information in short time, the sender produces a tag signal from spread frequency code and measures channel state information, then overlaps the tag signal on the communicational signal, the receiver can detect the correct tag signal and demodulate the communicational signal. This method avoids using complex cryptographic algorithm, reduces the amount of calculation, and prevents passive eavesdropping and active attack effectively. Simulation results show that this method has high use value.
2018, 40(5): 1072-1078.
doi: 10.11999/JEIT170735
Abstract:
In this paper, constacyclic codes over the finite fieldGF(q2) of length(q2m-1)/(q2-1) are studied. A sufficient and necessary condition for a class of constacyclic codes to be Hermitian dual-containing codes is given, and the parameters of this class of constacyclic codes are determined. Using Hermitian construction, the obtained quantum codes, are better than the parameters of quantum BCH codes.
In this paper, constacyclic codes over the finite fieldGF(q2) of length(q2m-1)/(q2-1) are studied. A sufficient and necessary condition for a class of constacyclic codes to be Hermitian dual-containing codes is given, and the parameters of this class of constacyclic codes are determined. Using Hermitian construction, the obtained quantum codes, are better than the parameters of quantum BCH codes.
2018, 40(5): 1079-1086.
doi: 10.11999/JEIT170712
Abstract:
Heterogeneous signcryption can be used to guarantee the confidentiality and the unforgeability in the different cryptographies. By analyzing some existing heterogeneous signcryption schemes, it is found that they only deal with a single message and can not achieve batch verification. Aggregation signcryption can not only take n distinct signcryption on n messages signed by n distinct users, but also provide a batch verification and reduce the cost of verification. In this paper, a Traditional Public Key Infrastructure (TPKI)-CertificateLess Public Key Cryptography (CLPKC) heterogeneous aggregation signcryption scheme is proposed, which can ensure the confidentiality and authentication between the TPKI and CLPKC. The scheme does not require bilinear pairings when it is aggregated. It is proved that the scheme has indistinguishability against adaptive chosen ciphertext attack and existential unforgeability against adaptive chosen messages attack under gap bilinear Diffie-Hellman and computational Diffie-Hellman problem and Discrete logarithm.
Heterogeneous signcryption can be used to guarantee the confidentiality and the unforgeability in the different cryptographies. By analyzing some existing heterogeneous signcryption schemes, it is found that they only deal with a single message and can not achieve batch verification. Aggregation signcryption can not only take n distinct signcryption on n messages signed by n distinct users, but also provide a batch verification and reduce the cost of verification. In this paper, a Traditional Public Key Infrastructure (TPKI)-CertificateLess Public Key Cryptography (CLPKC) heterogeneous aggregation signcryption scheme is proposed, which can ensure the confidentiality and authentication between the TPKI and CLPKC. The scheme does not require bilinear pairings when it is aggregated. It is proved that the scheme has indistinguishability against adaptive chosen ciphertext attack and existential unforgeability against adaptive chosen messages attack under gap bilinear Diffie-Hellman and computational Diffie-Hellman problem and Discrete logarithm.
2018, 40(5): 1087-1093.
doi: 10.11999/JEIT170730
Abstract:
The failure of a single physical server will cause a bad impact on the performance of the virtual networks in the cloud computing and data center environments. The existing approaches that provision redundant and backup physical resources are able to reduce the impact due to failure of physical devices, however, they do not pay attention to homogeneous issues of the physical servers. Thus, a heterogeneous-backup virtual network embedding method is proposed. Firstly, the redundant and backup physical resources are only provided to critical virtual machines so that the method can save the overhead of backup resources. Then, in order to improve resilience of the virtual networks, substrate nodes corresponding to the primary and backup embedding of each virtual machine must be heterogeneous. Finally, the total cost of provisioning bandwidth on the substrate links for the total virtual network is minimized as the objective function to further save the overhead of backup resources. Simulation experiments demonstrate that the proposed approach is able to significantly improve resilience of the virtual networks on the premise of guaranteeing the performance of virtual networks embedding.
The failure of a single physical server will cause a bad impact on the performance of the virtual networks in the cloud computing and data center environments. The existing approaches that provision redundant and backup physical resources are able to reduce the impact due to failure of physical devices, however, they do not pay attention to homogeneous issues of the physical servers. Thus, a heterogeneous-backup virtual network embedding method is proposed. Firstly, the redundant and backup physical resources are only provided to critical virtual machines so that the method can save the overhead of backup resources. Then, in order to improve resilience of the virtual networks, substrate nodes corresponding to the primary and backup embedding of each virtual machine must be heterogeneous. Finally, the total cost of provisioning bandwidth on the substrate links for the total virtual network is minimized as the objective function to further save the overhead of backup resources. Simulation experiments demonstrate that the proposed approach is able to significantly improve resilience of the virtual networks on the premise of guaranteeing the performance of virtual networks embedding.
2018, 40(5): 1094-1100.
doi: 10.11999/JEIT170617
Abstract:
To obtain better results for fine-grained video traffic classification, this paper analyzes the relationship between the feature variations during transmission and video traffic classification. According to the nature that different types of video services contain different downlink transmission rate variation patterns, a new video flow feature M value probability distribution, based on downlink byte rate variation is proposed, and video classification is realized by Support Vector Machine (SVM). The experimental results show that the probability distribution of M value is a better feature for classification of six kinds of common network video applications than other commonly used flow features.
To obtain better results for fine-grained video traffic classification, this paper analyzes the relationship between the feature variations during transmission and video traffic classification. According to the nature that different types of video services contain different downlink transmission rate variation patterns, a new video flow feature M value probability distribution, based on downlink byte rate variation is proposed, and video classification is realized by Support Vector Machine (SVM). The experimental results show that the probability distribution of M value is a better feature for classification of six kinds of common network video applications than other commonly used flow features.
2018, 40(5): 1101-1107.
doi: 10.11999/JEIT170787
Abstract:
A new network deployment algorithm is proposed for the problem of low network lifetime and low network coverage of underwater sensor networks. Firstly, the node which has a higher network coverage redundancy should be asleep. Then the network coverage and energy consumption will be set as the objective functions. And the multi-objective optimization algorithm will be adopted to optimize it. At last, the TOPSIS is used to select the best solution from the Non-dominated solution set. If any node is dead, the sleeping nodes in the near dead node will be waken up to preserve the coverage. The results demonstrate that the proposed algorithm outperform the existing algorithms in terms of various performance metrics including energy consumption and the coverage.
A new network deployment algorithm is proposed for the problem of low network lifetime and low network coverage of underwater sensor networks. Firstly, the node which has a higher network coverage redundancy should be asleep. Then the network coverage and energy consumption will be set as the objective functions. And the multi-objective optimization algorithm will be adopted to optimize it. At last, the TOPSIS is used to select the best solution from the Non-dominated solution set. If any node is dead, the sleeping nodes in the near dead node will be waken up to preserve the coverage. The results demonstrate that the proposed algorithm outperform the existing algorithms in terms of various performance metrics including energy consumption and the coverage.
2018, 40(5): 1108-1114.
doi: 10.11999/JEIT170590
Abstract:
The movement of the underwater gliders leads to the change in the position and relative distance among the gliders, which causes a change in the propagation delay of packets between gliders, and then it leads to a decrease in the reliability of communication between underwater gliders. The traditional underwater Media Access Control (MAC) protocols are for static topology networks and do not apply to dynamically varying network topology. Thus, a new MAC mechanism for the underwater glider networks is proposed. It predicts location based on underwater glider motion model. It calculates time slots dynamically according to the predicted results and the shared position information of underwater gliders. Then, it allocates time slots and reserves to the send and the receive. Underwater gliders avoid collisions with teamwork in the process of sending and receiving. Simulation results show that in this mechanism, the packet received rate increases by 12% and 25% comparing to the Prediction based MAC (P-MAC) protocol and Reservation based MAC (R-MAC) protocol, respectively. The result indicates that the new mechanism is more suitable for the dynamic network composed of underwater gliders.
The movement of the underwater gliders leads to the change in the position and relative distance among the gliders, which causes a change in the propagation delay of packets between gliders, and then it leads to a decrease in the reliability of communication between underwater gliders. The traditional underwater Media Access Control (MAC) protocols are for static topology networks and do not apply to dynamically varying network topology. Thus, a new MAC mechanism for the underwater glider networks is proposed. It predicts location based on underwater glider motion model. It calculates time slots dynamically according to the predicted results and the shared position information of underwater gliders. Then, it allocates time slots and reserves to the send and the receive. Underwater gliders avoid collisions with teamwork in the process of sending and receiving. Simulation results show that in this mechanism, the packet received rate increases by 12% and 25% comparing to the Prediction based MAC (P-MAC) protocol and Reservation based MAC (R-MAC) protocol, respectively. The result indicates that the new mechanism is more suitable for the dynamic network composed of underwater gliders.
2018, 40(5): 1115-1121.
doi: 10.11999/JEIT170710
Abstract:
Flat-topped beams can provide a uniform power density to the receiver in wireless power transmission, which is of great significance to improve the receiving efficiency and to simplify the design of rectifier circuit. This paper firstly studies the conditions of realizing flat-topped beam based on the Orchard synthesis by moving the zero points positions on the Schelkunoffs unit circle; secondly, it introduces the Sinc function and gives the relationship between the solution and the width of a flat-topped beam; finally, it discusses the relationship between the width and three parameters, rectangle coefficient, ripple factor, and side lobes level by using perturbation method and numerous calculations. An example of ten-element array is given to illustrate the procedure of achieving the narrowest flat-topped beam with a side lobe level less than verifying the efficiency of the methods and conclusion in this paper.
Flat-topped beams can provide a uniform power density to the receiver in wireless power transmission, which is of great significance to improve the receiving efficiency and to simplify the design of rectifier circuit. This paper firstly studies the conditions of realizing flat-topped beam based on the Orchard synthesis by moving the zero points positions on the Schelkunoffs unit circle; secondly, it introduces the Sinc function and gives the relationship between the solution and the width of a flat-topped beam; finally, it discusses the relationship between the width and three parameters, rectangle coefficient, ripple factor, and side lobes level by using perturbation method and numerous calculations. An example of ten-element array is given to illustrate the procedure of achieving the narrowest flat-topped beam with a side lobe level less than verifying the efficiency of the methods and conclusion in this paper.
2018, 40(5): 1122-1129.
doi: 10.11999/JEIT170721
Abstract:
A broadband planar printed Magneto-Electric (ME) dipole antenna and a Wide-Band Metamaterial Absorber (WBMA) with polarization-insensitive and wide-angle absorption are designed for X band application. The absorber exhibits absorptivity above 90% from 7.2 GHz to 12.6 GHz under normal incidence. When the incident angle increases to 45, the absorptivity is still above 90% at X band. By applying the absorber around the antenna, the RCS of antenna is reduced significantly. Simulated and experimental results show a monostatic RCS reduction above 3 dB from 7.2 GHz to 12.6 GHz, and the peak reduction is up to 23.8 dB under different polarized waves. At the center working frequency of 10 GHz, the bistatic RCS decreases significantly from-90 to +90 under TE polarized wave and from-35 to+35 under TM polarized wave while the radiation performance remains unchanged basically, which proves that the absorber has great absorptivity.
A broadband planar printed Magneto-Electric (ME) dipole antenna and a Wide-Band Metamaterial Absorber (WBMA) with polarization-insensitive and wide-angle absorption are designed for X band application. The absorber exhibits absorptivity above 90% from 7.2 GHz to 12.6 GHz under normal incidence. When the incident angle increases to 45, the absorptivity is still above 90% at X band. By applying the absorber around the antenna, the RCS of antenna is reduced significantly. Simulated and experimental results show a monostatic RCS reduction above 3 dB from 7.2 GHz to 12.6 GHz, and the peak reduction is up to 23.8 dB under different polarized waves. At the center working frequency of 10 GHz, the bistatic RCS decreases significantly from-90 to +90 under TE polarized wave and from-35 to+35 under TM polarized wave while the radiation performance remains unchanged basically, which proves that the absorber has great absorptivity.
2018, 40(5): 1130-1135.
doi: 10.11999/JEIT170793
Abstract:
To solve the technical problem of blind sorting of electronic interception signal, combined with the characteristics of pulse repetition interval stagger mode, based on the statistical clustering and data association, the algorithm for stagger pulse repetition interval pattern recognition is proposed, sequence analysis and signal sorting is carried out. The simulation results show that the proposed method can effectively dispose the blind signal sorting process and has strong robustness in case of missing and spurious pulses.
To solve the technical problem of blind sorting of electronic interception signal, combined with the characteristics of pulse repetition interval stagger mode, based on the statistical clustering and data association, the algorithm for stagger pulse repetition interval pattern recognition is proposed, sequence analysis and signal sorting is carried out. The simulation results show that the proposed method can effectively dispose the blind signal sorting process and has strong robustness in case of missing and spurious pulses.
2018, 40(5): 1136-1143.
doi: 10.11999/JEIT170707
Abstract:
High-resolution Direction Of Arrival (DOA) estimation is a critical issue for mainbeam multi-target tracking in ground-based or airborne early warning radar system. A Beam-Doppler Unitary ESPRIT (BD- UESPRIT) algorithm is proposed to deal with this problem. Firstly, multiple snapshots without spatial aperture loss are obtained using the technique of time-smoothing. Then the conjugate centrosymmetric Discrete Fourier Transform (DFT) matrix is used to transform the extracted data into beam-Doppler domain. Finally, the rotational invariance property of the space-time beam is exploited to estimate DOA. Since the proposed algorithm takes full advantage of temporal information and is implemented in low-dimensional beamspace, the DOA estimation accuracy can be improved greatly with dramatically reduced computational complexity. Numerical examples are given to verify the effectiveness of the proposed algorithm.
High-resolution Direction Of Arrival (DOA) estimation is a critical issue for mainbeam multi-target tracking in ground-based or airborne early warning radar system. A Beam-Doppler Unitary ESPRIT (BD- UESPRIT) algorithm is proposed to deal with this problem. Firstly, multiple snapshots without spatial aperture loss are obtained using the technique of time-smoothing. Then the conjugate centrosymmetric Discrete Fourier Transform (DFT) matrix is used to transform the extracted data into beam-Doppler domain. Finally, the rotational invariance property of the space-time beam is exploited to estimate DOA. Since the proposed algorithm takes full advantage of temporal information and is implemented in low-dimensional beamspace, the DOA estimation accuracy can be improved greatly with dramatically reduced computational complexity. Numerical examples are given to verify the effectiveness of the proposed algorithm.
2018, 40(5): 1144-1150.
doi: 10.11999/JEIT170723
Abstract:
This paper presents a new algorithm for self-calibration of spiral scanning airborne laser radar system. Firstly, the mathematic model of the point cloud location of the spiral scanning laser radar is deduced. Then the surface features of the overlapping area of the laser radar point cloud data is extracted, and the feature of the laser radar calibration model is constructed. Finally, the influence of the placement angle error on the forward and backward scanning data is analyzed, The plane parameters and the placement angle are estimated based on the regional network adjustment method by using the same name surface feature. The experimental results show that this method realizes the self-calibration of the placement angle of spiral scanning airborne laser radar system, and the absolute accuracy satisfies requirement of the production data.
This paper presents a new algorithm for self-calibration of spiral scanning airborne laser radar system. Firstly, the mathematic model of the point cloud location of the spiral scanning laser radar is deduced. Then the surface features of the overlapping area of the laser radar point cloud data is extracted, and the feature of the laser radar calibration model is constructed. Finally, the influence of the placement angle error on the forward and backward scanning data is analyzed, The plane parameters and the placement angle are estimated based on the regional network adjustment method by using the same name surface feature. The experimental results show that this method realizes the self-calibration of the placement angle of spiral scanning airborne laser radar system, and the absolute accuracy satisfies requirement of the production data.
2018, 40(5): 1151-1158.
doi: 10.11999/JEIT170426
Abstract:
Due to the large time-bandwidth product, Linear Frequency Modulation (LFM) signals are widely used, and their diversity can be applied to orthogonal waveform set design for MIMO radar. To solve the problems of the correlation functions of the existing waveforms, detailed analysis is made, a new waveform based on chirp durations is proposed, and the orthogonal waveform set design method is given subsequently. Peak sidelobe level is calculated for the cost function, the chirp durations of subpulses are optimized by Sequential Quadratic Programming (SQP). Simulation results show that the designed waveforms have lower autocorrelation sidelobe level and crosscorrelation level compared with the present method. In addition, relationships among autocorrelation sidelobe level, crosscorrelation level and the size of orthogonal waveforms are studied via numerical experiments.
Due to the large time-bandwidth product, Linear Frequency Modulation (LFM) signals are widely used, and their diversity can be applied to orthogonal waveform set design for MIMO radar. To solve the problems of the correlation functions of the existing waveforms, detailed analysis is made, a new waveform based on chirp durations is proposed, and the orthogonal waveform set design method is given subsequently. Peak sidelobe level is calculated for the cost function, the chirp durations of subpulses are optimized by Sequential Quadratic Programming (SQP). Simulation results show that the designed waveforms have lower autocorrelation sidelobe level and crosscorrelation level compared with the present method. In addition, relationships among autocorrelation sidelobe level, crosscorrelation level and the size of orthogonal waveforms are studied via numerical experiments.
2018, 40(5): 1159-1165.
doi: 10.11999/JEIT170775
Abstract:
This paper proposes an improved multi-baseline InSAR height reconstruction approach to reconstruct height models of discontinuous terrains through combining with the Maximum A Posteriori (MAP) estimation based on Total Variation (TV) model energy function and a defective pixel judgment method composed of interferogram residues and comparison of adjacent pixels differences. The proposed method utilizes the difference discontinuous pixels between interferogram residues and comparison of adjacent pixels differences, to make defective pixels judgment from reconstructed terrain graphs obtained by the MAP estimation based on TV model energy function, which estimates efficiently defective pixels caused by errors or noise from discontinuous pixels, and then updates relatively defective pixels to obtain the final terrain height estimation. The presented algorithm not only keeps the convenience of MAP estimation algorithm but also improves estimation precision for height models of objective terrains. Experiences of two different types of terrain show that the proposed approach is efficient as well as valid.
This paper proposes an improved multi-baseline InSAR height reconstruction approach to reconstruct height models of discontinuous terrains through combining with the Maximum A Posteriori (MAP) estimation based on Total Variation (TV) model energy function and a defective pixel judgment method composed of interferogram residues and comparison of adjacent pixels differences. The proposed method utilizes the difference discontinuous pixels between interferogram residues and comparison of adjacent pixels differences, to make defective pixels judgment from reconstructed terrain graphs obtained by the MAP estimation based on TV model energy function, which estimates efficiently defective pixels caused by errors or noise from discontinuous pixels, and then updates relatively defective pixels to obtain the final terrain height estimation. The presented algorithm not only keeps the convenience of MAP estimation algorithm but also improves estimation precision for height models of objective terrains. Experiences of two different types of terrain show that the proposed approach is efficient as well as valid.
2018, 40(5): 1166-1172.
doi: 10.11999/JEIT170806
Abstract:
Researchers generally consider that Ratio Of Exponentially Weighted Averages (ROEWA) can not calculate the edge directions of SAR images. Therefore, some directional filters are used to add directions to ROEWA. In this paper, an Enhanced ROEWA (EROEWA) algorithm is proposed. Through the further derivation of the ROEWA algorithm convolution process, the pixel-level observation formula of ROEWA algorithm is obtained. First, a new convolution strategy is used to decouple the ROEWA formula to obtain exponentially weighted averages over the four directions. Second, the SAR images are rotated by 45 and exponentially weighted averages of the four additional directions are calculated. Finally, exponentially weighted averages of eight directions are expressed as eight vectors, and edge intensity and direction are solved by vector synthesis. Experimental results show that the EROEWA has a significant enhancement effect on the edge intensity and the direction.
Researchers generally consider that Ratio Of Exponentially Weighted Averages (ROEWA) can not calculate the edge directions of SAR images. Therefore, some directional filters are used to add directions to ROEWA. In this paper, an Enhanced ROEWA (EROEWA) algorithm is proposed. Through the further derivation of the ROEWA algorithm convolution process, the pixel-level observation formula of ROEWA algorithm is obtained. First, a new convolution strategy is used to decouple the ROEWA formula to obtain exponentially weighted averages over the four directions. Second, the SAR images are rotated by 45 and exponentially weighted averages of the four additional directions are calculated. Finally, exponentially weighted averages of eight directions are expressed as eight vectors, and edge intensity and direction are solved by vector synthesis. Experimental results show that the EROEWA has a significant enhancement effect on the edge intensity and the direction.
2018, 40(5): 1173-1180.
doi: 10.11999/JEIT170737
Abstract:
The performance of direction finding for nested array degrades due to the mutual coupling effect among the elements. Two different translational nested array structures are proposed. In order to ensure that the virtual array has no holes, a translational nested array is formed by adjusting the positions of the original two level nested array elements. It improves the sparsity of the original two level nested array, reduces the mutual coupling effect, and extends the direction finding freedom of the original nested array. Under the condition of unknown number of spatial radiation sources, a Sparse Bayesian Learning (SBL) model for translational nested array is established. Through this model, the received data of the virtual array is processed, the DOA estimation is obtained and the direction finding performance of the original nested array direction finding algorithm is effectively improved. Simulation results show that the translational nested array has higher degree of freedom than the original nested array. Under the scenarios of low Signal-to-Noise Ratio (SNR), snapshot deficiency, and mutual coupling effect, the performance of direction finding algorithm for translational nested array based on Sparse Bayesian Learning is better than that of direction finding algorithm for the original nested array. The angle resolution of direction finding algorithm for the original nested array is improved.
The performance of direction finding for nested array degrades due to the mutual coupling effect among the elements. Two different translational nested array structures are proposed. In order to ensure that the virtual array has no holes, a translational nested array is formed by adjusting the positions of the original two level nested array elements. It improves the sparsity of the original two level nested array, reduces the mutual coupling effect, and extends the direction finding freedom of the original nested array. Under the condition of unknown number of spatial radiation sources, a Sparse Bayesian Learning (SBL) model for translational nested array is established. Through this model, the received data of the virtual array is processed, the DOA estimation is obtained and the direction finding performance of the original nested array direction finding algorithm is effectively improved. Simulation results show that the translational nested array has higher degree of freedom than the original nested array. Under the scenarios of low Signal-to-Noise Ratio (SNR), snapshot deficiency, and mutual coupling effect, the performance of direction finding algorithm for translational nested array based on Sparse Bayesian Learning is better than that of direction finding algorithm for the original nested array. The angle resolution of direction finding algorithm for the original nested array is improved.
2018, 40(5): 1181-1186.
doi: 10.11999/JEIT170727
Abstract:
In order to exploit the development potential of current measurement and control equipment and to build an omnibus underwater measurement and control network with higher precision, according to the working characteristics of the system itself, an underwater measurement and control network centralized data fusion localization algorithm based on Chan-algorithm method is proposed. It is an algorithm that coarsely calculates the target position according to the time of arrival by using the Weighted Least Squares (WLS) estimation method at first; secondly, it constructs new error vectors on the basis of the relationship between the position and the time delay information; last, the target position is resolved from the vectors by using WLS method again. The result of research demonstrates that the algorithm realizes the data fusion of multiple sets of underwater measurement and control equipment, which could improve the precision of localization globally. And the precision of the proposed method is much better than the data fusion localization algorithm purely based on the time of arrival.
In order to exploit the development potential of current measurement and control equipment and to build an omnibus underwater measurement and control network with higher precision, according to the working characteristics of the system itself, an underwater measurement and control network centralized data fusion localization algorithm based on Chan-algorithm method is proposed. It is an algorithm that coarsely calculates the target position according to the time of arrival by using the Weighted Least Squares (WLS) estimation method at first; secondly, it constructs new error vectors on the basis of the relationship between the position and the time delay information; last, the target position is resolved from the vectors by using WLS method again. The result of research demonstrates that the algorithm realizes the data fusion of multiple sets of underwater measurement and control equipment, which could improve the precision of localization globally. And the precision of the proposed method is much better than the data fusion localization algorithm purely based on the time of arrival.
2018, 40(5): 1187-1194.
doi: 10.11999/JEIT170758
Abstract:
In order to improve the performance of speech communication system in noisy environment, a dual micro-array speech enhancement method based on subband spectrum subtraction and generalized sidelobe canceller is proposed. Based on dual micro-array structure and the subband structure analysis, this method suppresses the noise in the low frequency band by the spectrum subtraction method which may employ variable over-subtraction factor in it and suppresses the non-coherent noise part in the high frequency band by modifying cross power spectrum subtraction method, respectively. Then generalized sidelobe canceller and voice activity detector are combined with the method mentioned above to further suppress the influence of strong coherent noise further. The experimental results show that the proposed method can suppress the influence of noise and improve the intelligibility of speech more effectively.
In order to improve the performance of speech communication system in noisy environment, a dual micro-array speech enhancement method based on subband spectrum subtraction and generalized sidelobe canceller is proposed. Based on dual micro-array structure and the subband structure analysis, this method suppresses the noise in the low frequency band by the spectrum subtraction method which may employ variable over-subtraction factor in it and suppresses the non-coherent noise part in the high frequency band by modifying cross power spectrum subtraction method, respectively. Then generalized sidelobe canceller and voice activity detector are combined with the method mentioned above to further suppress the influence of strong coherent noise further. The experimental results show that the proposed method can suppress the influence of noise and improve the intelligibility of speech more effectively.
2018, 40(5): 1195-1201.
doi: 10.11999/JEIT170624
Abstract:
Accurate and reliable blind noise estimation is an important research topic of digital image processing. The main challenge is how to extract pure noise information for estimating. In recent years, many algorithms use principal component analysis technology to exclude the interference of image textures information, and estimate noise level by using the minimal eigenvalue. So that, the image textures have smallest effect on the minimal eigenvalue, thus this kind of methods performs well for high frequency image (image with abundant textures). The minimal eigenvalue is actually smaller than the true noise variance because of limited image blocks, and the bias is the bigger if the number of image patches is the smaller. If the noise level is estimated as the smallest eigenvalue, the final result will be underestimated. It is found that the relation between the ratio of estimated result to real noise variance and the number of image blocks is power function by using regression analysis, thus the true noise level can be computed by using the minimal eigenvalue and the power function. The experiment results show that the proposed algorithm works well over a large range of visual content and noise conditions, and can process multiply Gaussian noise too.
Accurate and reliable blind noise estimation is an important research topic of digital image processing. The main challenge is how to extract pure noise information for estimating. In recent years, many algorithms use principal component analysis technology to exclude the interference of image textures information, and estimate noise level by using the minimal eigenvalue. So that, the image textures have smallest effect on the minimal eigenvalue, thus this kind of methods performs well for high frequency image (image with abundant textures). The minimal eigenvalue is actually smaller than the true noise variance because of limited image blocks, and the bias is the bigger if the number of image patches is the smaller. If the noise level is estimated as the smallest eigenvalue, the final result will be underestimated. It is found that the relation between the ratio of estimated result to real noise variance and the number of image blocks is power function by using regression analysis, thus the true noise level can be computed by using the minimal eigenvalue and the power function. The experiment results show that the proposed algorithm works well over a large range of visual content and noise conditions, and can process multiply Gaussian noise too.
2018, 40(5): 1202-1209.
doi: 10.11999/JEIT170717
Abstract:
Most of the tracking-by-detection algorithms treat the tracking task as a category classification task, when the target experience deformation or encounter similar objects interference, the model drift is prone to occur. In this paper, a multi-exemplar regression tracking algorithm is proposed. In this algorithm, the exemplar model is considered to be more appropriate for tracking task, the exemplar model is set up by a frame image information, and the multi-exemplar model established in the time series can represent the target current state; in order to make the tracking algorithm adapt to the target deformation, the exemplar model is considered as the hidden variable by logistic regression model, together with the training sets from several recent frames sampling, can jointly build multi-exemplar regression tracking model. As the tracker builds multi-exemplar model on the whole, linking them together closely, it can effectively deal with the target deformation. Since the model drift only affects the exemplar model at current frame, each exemplar model is independent of each other, so the tracking algorithm can effectively reduce the influence of model drift on robust tracking. In the experiment, OTB 2013 benchmark and UAV 123 benchmark are used to verify the algorithm, DeepSRDCF, Siamese-fc and other algorithms act as the contrast algorithms, the experimental results show that the proposed tracker not only gives full play to the advantages of tracking based on multi-exemplar regression model, but also has good performance in deformation and background blur scene, and achieves three to five percent more than other advanced algorithms in the metrics of success rate and precision.
Most of the tracking-by-detection algorithms treat the tracking task as a category classification task, when the target experience deformation or encounter similar objects interference, the model drift is prone to occur. In this paper, a multi-exemplar regression tracking algorithm is proposed. In this algorithm, the exemplar model is considered to be more appropriate for tracking task, the exemplar model is set up by a frame image information, and the multi-exemplar model established in the time series can represent the target current state; in order to make the tracking algorithm adapt to the target deformation, the exemplar model is considered as the hidden variable by logistic regression model, together with the training sets from several recent frames sampling, can jointly build multi-exemplar regression tracking model. As the tracker builds multi-exemplar model on the whole, linking them together closely, it can effectively deal with the target deformation. Since the model drift only affects the exemplar model at current frame, each exemplar model is independent of each other, so the tracking algorithm can effectively reduce the influence of model drift on robust tracking. In the experiment, OTB 2013 benchmark and UAV 123 benchmark are used to verify the algorithm, DeepSRDCF, Siamese-fc and other algorithms act as the contrast algorithms, the experimental results show that the proposed tracker not only gives full play to the advantages of tracking based on multi-exemplar regression model, but also has good performance in deformation and background blur scene, and achieves three to five percent more than other advanced algorithms in the metrics of success rate and precision.
2018, 40(5): 1210-1218.
doi: 10.11999/JEIT170827
Abstract:
Most existing salient object detection methods based on diffusion theory usually only use one feature of image to construct graph and diffusion matrix, and ignore the possibility that salient objects appear at the border regions of the image. In this paper, a diffusion method based on the multi-layer features of image is proposed to detect salient objects. Firstly, the seed nodes are selected by adopting the high-level prior method, which is composed of background prior, color prior, and location prior. Then, the initial saliency map is obtained by propagating the saliency information carried by the selected seed nodes to each nodes via the diffusion matrix constructed by the low-level feature of the image, and used as the middle-level feature of image. The diffusion matrices are re-synthesized again by the middle-level feature and the high-level feature of the image, and the middle-level saliency map and the high-level saliency map are obtained by the diffusion-based method respectively. The final saliency map is obtained by nonlinearly combining the the middle-level and high-level saliency map. Results on three datasets, MSRA10K, DUT-OMRON and ECSSD, show that the proposed method achieves superior performance compared with the four state-of-art methods in terms of three evaluation metrics.
Most existing salient object detection methods based on diffusion theory usually only use one feature of image to construct graph and diffusion matrix, and ignore the possibility that salient objects appear at the border regions of the image. In this paper, a diffusion method based on the multi-layer features of image is proposed to detect salient objects. Firstly, the seed nodes are selected by adopting the high-level prior method, which is composed of background prior, color prior, and location prior. Then, the initial saliency map is obtained by propagating the saliency information carried by the selected seed nodes to each nodes via the diffusion matrix constructed by the low-level feature of the image, and used as the middle-level feature of image. The diffusion matrices are re-synthesized again by the middle-level feature and the high-level feature of the image, and the middle-level saliency map and the high-level saliency map are obtained by the diffusion-based method respectively. The final saliency map is obtained by nonlinearly combining the the middle-level and high-level saliency map. Results on three datasets, MSRA10K, DUT-OMRON and ECSSD, show that the proposed method achieves superior performance compared with the four state-of-art methods in terms of three evaluation metrics.
2018, 40(5): 1219-1225.
doi: 10.11999/JEIT170608
Abstract:
The Direct Position Determination (DPD) methods outperform the classic two-step localization methods in localization accuracy, while the existing DPD methods based on array antenna models do not exploit Doppler frequency shift information to improve localization accuracy further. Fusing Doppler frequency shift information, a DPD method based on array antenna model is proposed to overcome the above mentioned shortcoming. Firstly, a DPD model fusing Doppler frequency shift information is constructed. Then, a maximum likelihood estimator is designed and emitters position estimation is transformed to calculating the maximum eigenvalue of the matrix containing location information. The property of eigenvalue keeping unchanged after matrix transpose is used to simplify the calculation. Finally, the emitters position is estimated via a two-dimensional grid search. Simulations show that the proposed method is superior over the DPD methods based on array antenna models and the DPD methods only using Doppler frequency shift information in localization accuracy.
The Direct Position Determination (DPD) methods outperform the classic two-step localization methods in localization accuracy, while the existing DPD methods based on array antenna models do not exploit Doppler frequency shift information to improve localization accuracy further. Fusing Doppler frequency shift information, a DPD method based on array antenna model is proposed to overcome the above mentioned shortcoming. Firstly, a DPD model fusing Doppler frequency shift information is constructed. Then, a maximum likelihood estimator is designed and emitters position estimation is transformed to calculating the maximum eigenvalue of the matrix containing location information. The property of eigenvalue keeping unchanged after matrix transpose is used to simplify the calculation. Finally, the emitters position is estimated via a two-dimensional grid search. Simulations show that the proposed method is superior over the DPD methods based on array antenna models and the DPD methods only using Doppler frequency shift information in localization accuracy.
2018, 40(5): 1226-1233.
doi: 10.11999/JEIT170663
Abstract:
In the Direction Of Arrival (DOA) estimation of incoherently distributed noncircular sources, the increase of dimension caused by array output matrix extension can cause a large computational complexity. To solve this problem, a rapid DOA estimation algorithm is proposed based on cross-correlation sampling decomposition. It only needs to calculate two low-dimensional sub-matrices, which are formed by a small number of rows and columns in the extended Cross-Correlation (CC) matrix. On the premise of the sub-matrices, the right and left singular vectors corresponding to two signal subspaces can be simultaneously obtained by the low-rank approximation decomposition, which avoids the calculation of the whole covariance matrix and its singular value decomposition. Finally, the DOA estimation can be obtained by the least squares with the rotation invariance of the signal subspaces. The simulation results show that when the number of samples in the low-dimensional sub-matrix is larger than twice the number of sources, the performance of the proposed algorithm is comparable with the DOA estimation algorithm of incoherently distributed noncircular sources based on the singular value decomposition applying to the CC matrix. Moreover, the proposed algorithm utilizes the noncircular characteristic of the signal to achieve higher estimation performance compared with the traditional low-complexity DOA estimation algorithms of the incoherently distributed sources.
In the Direction Of Arrival (DOA) estimation of incoherently distributed noncircular sources, the increase of dimension caused by array output matrix extension can cause a large computational complexity. To solve this problem, a rapid DOA estimation algorithm is proposed based on cross-correlation sampling decomposition. It only needs to calculate two low-dimensional sub-matrices, which are formed by a small number of rows and columns in the extended Cross-Correlation (CC) matrix. On the premise of the sub-matrices, the right and left singular vectors corresponding to two signal subspaces can be simultaneously obtained by the low-rank approximation decomposition, which avoids the calculation of the whole covariance matrix and its singular value decomposition. Finally, the DOA estimation can be obtained by the least squares with the rotation invariance of the signal subspaces. The simulation results show that when the number of samples in the low-dimensional sub-matrix is larger than twice the number of sources, the performance of the proposed algorithm is comparable with the DOA estimation algorithm of incoherently distributed noncircular sources based on the singular value decomposition applying to the CC matrix. Moreover, the proposed algorithm utilizes the noncircular characteristic of the signal to achieve higher estimation performance compared with the traditional low-complexity DOA estimation algorithms of the incoherently distributed sources.
2018, 40(5): 1234-1241.
doi: 10.11999/JEIT170675
Abstract:
The performance of the K2 algorithm depends on node ordering heavily, and the genetic algorithm can not find the node ordering effectively. For these problems, a new Bayesian structure learning algorithm, named NOK2 (Node Ordering searching for K2 algorithm), is proposed to solve the Bayesian structure learning problem by searching node ordering directly. According to the requirements of K2 algorithm based on prior knowledge and the weight matrix of spanning tree, the fitness function for quantitative evaluation of node ordering is established. The genetic algorithm is redesigned by a new method combines the dynamic learning constants, the hybrid crossover strategy, the inverted mutation strategy and the isolated node processing, so that the algorithm can find the node order of the highest fitness value, and this node sequence is taken as a prior knowledge of the K2 algorithm to obtain the optimal Bayesian network structure. Compared with other optimization algorithms, experimental results indicate that the NOK2 algorithm can significantly increase nearly 13.11% in the scoring metric values.
The performance of the K2 algorithm depends on node ordering heavily, and the genetic algorithm can not find the node ordering effectively. For these problems, a new Bayesian structure learning algorithm, named NOK2 (Node Ordering searching for K2 algorithm), is proposed to solve the Bayesian structure learning problem by searching node ordering directly. According to the requirements of K2 algorithm based on prior knowledge and the weight matrix of spanning tree, the fitness function for quantitative evaluation of node ordering is established. The genetic algorithm is redesigned by a new method combines the dynamic learning constants, the hybrid crossover strategy, the inverted mutation strategy and the isolated node processing, so that the algorithm can find the node order of the highest fitness value, and this node sequence is taken as a prior knowledge of the K2 algorithm to obtain the optimal Bayesian network structure. Compared with other optimization algorithms, experimental results indicate that the NOK2 algorithm can significantly increase nearly 13.11% in the scoring metric values.
2018, 40(5): 1242-1249.
doi: 10.11999/JEIT170745
Abstract:
For object detection in fisheye images which present serious distortion, an object tracking method is proposed to deal with scale variance, pose change and distortion. Firstly, gray feature and gradient feature are combined to obtain a high dimensional feature of the target, then reduce its dimensionality by averaging to obtain targets compressive feature. According to fisheye imaging model, motion of object point is modeled, and range of motion of target is predicted. In order to adjust to scale variance, corner points are positioned respectively in a coarse to fine manner based on the block matching motion estimation, and the scale of compressed feature is changed along with scale change of object box. Experimental results show that the proposed algorithm is superior to other algorithms in the case of distortion, scale change, pose change and part occlusion.
For object detection in fisheye images which present serious distortion, an object tracking method is proposed to deal with scale variance, pose change and distortion. Firstly, gray feature and gradient feature are combined to obtain a high dimensional feature of the target, then reduce its dimensionality by averaging to obtain targets compressive feature. According to fisheye imaging model, motion of object point is modeled, and range of motion of target is predicted. In order to adjust to scale variance, corner points are positioned respectively in a coarse to fine manner based on the block matching motion estimation, and the scale of compressed feature is changed along with scale change of object box. Experimental results show that the proposed algorithm is superior to other algorithms in the case of distortion, scale change, pose change and part occlusion.
2018, 40(5): 1250-1257.
doi: 10.11999/JEIT170794
Abstract:
In order to improve quality of the reconstructed images of the Electrical Capacitance Tomography (ECT) system, an improved sparsity adaptive matching pursuit compressed sensing algorithm is proposed. Based on the coherence point of Compressed Sensing (CS) theory and ECT, the CS-ECT model is established. In the model, the sensitivity matrix of ECT is designed in a random order to be the observation matrix, the discrete cosine base is used as the sparse base, the capacitance value is measured as the observed value. By using the Linear Back Projection (LBP) algorithm, the sparsity of the estimated images is confirmed. The sparsity can be served as the initial value of the atomic index for sparsity adaptive iteration. The lack of image reconstruction accuracy caused by the inaccurate estimate of sparsity can be solved by the improved sparsity adaptive matching pursuit algorithm. Simulation results indicate that reconstructed images with higher accuracy can be obtained using the improved sparsity adaptive matching pursuit compressed sensing algorithm than the LBP algorithm, Landweber algorithm and Tikhonov algorithm. A new method of ECT reconstruction is provided.
In order to improve quality of the reconstructed images of the Electrical Capacitance Tomography (ECT) system, an improved sparsity adaptive matching pursuit compressed sensing algorithm is proposed. Based on the coherence point of Compressed Sensing (CS) theory and ECT, the CS-ECT model is established. In the model, the sensitivity matrix of ECT is designed in a random order to be the observation matrix, the discrete cosine base is used as the sparse base, the capacitance value is measured as the observed value. By using the Linear Back Projection (LBP) algorithm, the sparsity of the estimated images is confirmed. The sparsity can be served as the initial value of the atomic index for sparsity adaptive iteration. The lack of image reconstruction accuracy caused by the inaccurate estimate of sparsity can be solved by the improved sparsity adaptive matching pursuit algorithm. Simulation results indicate that reconstructed images with higher accuracy can be obtained using the improved sparsity adaptive matching pursuit compressed sensing algorithm than the LBP algorithm, Landweber algorithm and Tikhonov algorithm. A new method of ECT reconstruction is provided.
2018, 40(5): 1258-1265.
doi: 10.11999/JEIT170815
Abstract:
In Chinese text categorization tasks, the locations of the important features in the Chinese texts are disperse and sparse, and the different characteristics of Chinese texts contributes differently for the recognition of their categories. In order to solve the above problems, this paper proposes a multi-feature fusion model Three Convolutional neural network paths and Long short term memory path fused with Attention neural network path (3CLA) for Chinese text categorization, which is based on Convolutional Neural Network (CNN), Long Short Term Memory (LSTM) and semantic understanding attention neural networks. The model first uses text preprocessing to finish the segmentation and vectorization of the Chinese text. Then, through the embedding layer, the input data are sent to the CNN path, the LSTM path and the attention path respectively to extract text features of different levels and different characteristics. Finally, the text features are fused by the fusion layer and classified by the classifier. Based on the Chinese corpus, the text classification experiment is carried out. The results of the experiments show that compared with the CNN structure model and the LSTM structure model, the proposed algorithm model improves the recognition ability of Chinese text categories by up to about 8%.
In Chinese text categorization tasks, the locations of the important features in the Chinese texts are disperse and sparse, and the different characteristics of Chinese texts contributes differently for the recognition of their categories. In order to solve the above problems, this paper proposes a multi-feature fusion model Three Convolutional neural network paths and Long short term memory path fused with Attention neural network path (3CLA) for Chinese text categorization, which is based on Convolutional Neural Network (CNN), Long Short Term Memory (LSTM) and semantic understanding attention neural networks. The model first uses text preprocessing to finish the segmentation and vectorization of the Chinese text. Then, through the embedding layer, the input data are sent to the CNN path, the LSTM path and the attention path respectively to extract text features of different levels and different characteristics. Finally, the text features are fused by the fusion layer and classified by the classifier. Based on the Chinese corpus, the text classification experiment is carried out. The results of the experiments show that compared with the CNN structure model and the LSTM structure model, the proposed algorithm model improves the recognition ability of Chinese text categories by up to about 8%.
2018, 40(5): 1266-1270.
doi: 10.11999/JEIT170461
Abstract:
A miniaturized high resolution W-band SAR system developed by the Institute of Electronics of the Chinese Academy of Sciences is introduced. The parameters, architecture of the system and signal processing method are described in detail. The solutions of key problems which include non-linearities error calibration, high-accuracy motion compensation and high-isolation transceiver system design are proposed. The W-band radar system is tested by airborne experiment and the results show the effectiveness of the presented system.
A miniaturized high resolution W-band SAR system developed by the Institute of Electronics of the Chinese Academy of Sciences is introduced. The parameters, architecture of the system and signal processing method are described in detail. The solutions of key problems which include non-linearities error calibration, high-accuracy motion compensation and high-isolation transceiver system design are proposed. The W-band radar system is tested by airborne experiment and the results show the effectiveness of the presented system.
