一种基于PCNN赋时矩阵的图像去噪新算法

刘勍; 马义德

doi:10.3724/SP.J.1146.2006.02059

一种基于PCNN赋时矩阵的图像去噪新算法

doi: 10.3724/SP.J.1146.2006.02059

刘勍,
马义德

基金项目:

国家自然科学基金(60572011)和甘肃省教育厅科研项目基金(0708- 10)资助课题

计量
- 文章访问数: 3099
- HTML全文浏览量: 74
- PDF下载量: 870
- 被引次数: 0
出版历程
- 收稿日期: 2006-12-26
- 修回日期: 2007-11-23
- 刊出日期: 2008-08-19

A New Algorithm for Noise Reducing of Image Based on PCNN Time Matrix

Liu Qing,
Ma Yi-de

摘要

摘要: 该文从图像脉冲噪声的特点出发，提出了基于脉冲耦合神经网络(Pulse Coupled Neural Networks, PCNN)赋时矩阵的图像去噪算法。赋时矩阵是由PCNN产生的一种从空间图像信息到时间信息的映射图，在图像处理中，赋时矩阵包含有与空间相联系的有用信息。计算机仿真结果表明，通过对PCNN赋时矩阵分析与处理，综合运用相关方法，可以有效地滤除被脉冲噪声污染的图像噪声，且恢复图像的视觉效果明显地好于中值滤波、均值滤波及维纳法得到的结果，其信噪比高、去噪能力强、对边缘和细节的保护性好、适应性强。
- 图像去噪; PCNN; 赋时矩阵; 脉冲噪声
Abstract: This paper puts forward a new algorithm for image reducing noise based on the time matrix of Pulse Coupled Neural Networks (PCNN) form the aspect of the characteristic of image impulsive noise. The time matrix is a mapping from spatial image information to time information generated from PCNN, The time matrix contains useful information related to spatial information in image processing. The results of computer simulations show that through analyzing and processing the PCNN time matrix, the image polluted by impulsive noise can be filtered efficiently and visual effects of restoration images are much batter than those obtained from the median filter, mean filter and wiener filter. This method presents higher Peak Signal-to-Noise, better capability to reduce noise, and can protect edges and details of images and be more adaptive.

HTML全文

参考文献(1)

Yin L and Neuvo Y. Adaptive FIR-WOS hybrid filtering [A].Proc. Int. Symp. Circuits and Syst [C]. San Diego, CA, 1992,6: 2637-2640.[2]Eng How-Lung and Ma Kai-Kuang. Noise adaptive sofetswitchingmedian filter[J].IEEE Trans. on Image Processing.2001, 10(2):242-251[3]Wang J H and Yu M D. Images smoothing by adaptive fuzzyoptimal filter[A].Proc. IEEE Int.Conf.Syst. Man and Cybern[C]. Vancouver, BC, Canada, 1995: 845-848.[4]Wang J H and Lin L D. An improved median filter usingminmax algorithm for image processing [J].Electron Lett.1997, 33(16):1362-1363[5]Sun T and Neuvo Y. Detail-preserving median based filters inimage processing [J].Pattern Recognition Letter.1994, 15(4):341-347[6]Eckhorn R, Reitboeck H J, and Arndtetal M. Feature linkingvia synchronization among distributed assemblies: simulationof results from cat cortex [J].Neural Computation.1990, 2(3):293-307[7]Ranganath H S, Kuntimad G, and Johnson J L.Pulse-coupled neural network for image processing [A]. In:Proceedings of IEEE Southeastcon [C]. New York: IEEEPress, 1995: 37-43.[8]Johnson J L and Padgett M L. PCNN models andapplications [J].IEEE Trans. on Neural Networks.1999,10(3):480-498[9]Kuntimad G and Ranganath H S. Perfect imagesegmentation using pulse coupled neural networks [J].IEEETrans. on Neural Networks.1999, 10(3):591-598[10]Ranganath H S and Kuntimad G. Object detection usingpulse coupled neural networks[J]. IEEE Trans. on NeuralNetworks, 1999, 10(3): 615-620.[11]刘勍, 马义德, 钱志柏.一种基于交叉熵的改进型PCNN 图像自动分割新方法[J]. 中国图象图形学报, 2005, 10(5): 579-584.Liu Q, Ma Y D, and Qian Z B. Automated imagesegmentation using improved PCNN model based oncross-entropy [J]. Journal of Image and Graphics, 2005, 10(5):579-584.[12]Chacon M I and Zimmerman A. PCNNP: A pulse-coupledneural network processor [A]. IEEE International Joint.Conference on Neural Networks [C]. Honolulu Hawaii USA.2002: 1581-1584.[13]Stewart R D, Fermin I, and Opper M. Region growing withpulse-coupled neural networks: an alternative to seededregion growing [J].IEEE Trans. on Neural Network.2002,13(6):1557-1562[14]Ma Y D, Dai R L, and Li L. Image segmentation of embryonicplant cell using pulse-coupled neural networks [J]. ChineseScience Bulletin, 2002, 47(2): 167-172.

施引文献

资源附件(0)

访问统计