基于多尺度分解的k邻域随机查找快速图像修复

廖斌; 苏涛; 刘斌

doi:10.11999/JEIT150033

基于多尺度分解的k邻域随机查找快速图像修复

doi: 10.11999/JEIT150033

廖斌¹,
苏涛¹,
刘斌¹

基金项目:

国家自然科学基金(61300125, 61471160)

计量
- 文章访问数: 1423
- HTML全文浏览量: 96
- PDF下载量: 707
- 被引次数: 0
出版历程
- 收稿日期: 2015-01-06
- 修回日期: 2015-04-08
- 刊出日期: 2015-09-19

Multi-scale Decomposition Based k-nearest-neighbor Random Search for Fast Image Completion

摘要

摘要: 该文提出一种基于多尺度分解的k邻域随机查找快速图像修复方法。基于双边滤波下采样分解图像，从图像最粗糙层开始，对每一粗糙层采用基于最小堆的k邻域随机查找算法快速搜索最佳匹配块，利用鲁棒优先级函数确定下一待修复块。每一粗糙层修复后用双边滤波上采样重建下一粗糙层，迭代得到最终的修复结果。与相关工作比较，所提方法的修复结果能够保持图像的细节和边缘信息，取得更高的修复质量。利用客观指标评价修复结果。实验结果表明该方法有效易行，修复的图像具有良好的可视效果。
- 图像处理 /
- 图像修复 /
- 多尺度分解 /
- k邻域随机查找
Abstract: Multi-scale decomposition based k-nearest-neighbor random search for fast image completion is presented. The image is decomposed using the bilateral filtering based down sampling. Starting from the coarsest level image, the most matching patch is searched using k-nearest-neighbor search algorithm based on the minimum heap for each coarse layer. The robust priority function is presented to determine the next patch that should be handled. The lower coarse layer is reconstructed using the bilateral filtering based up sampling after current coarse layer is repaired, so as to get the final result with iterative completion. Compared with related work, the presented algorithm preserves image details and edge information, and obtains higher completion quality. The completion results are evaluated utilizing the objective indictors. The experimental results show that presented method is effective, feasible, and the visual effect of the image completion is pleasing.
- Image processing /
- Image completion /
- Multi-scale decomposition /
- k-nearest-neighbor random search

HTML全文

参考文献(18)

He Kai-ming and Sun Jian. Image completion approaches using the statistics of similar patches[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2014, 36(12): 2423-2435.

许建楼, 冯象初, 郝岩. 改进的TV-Stokes图像修复模型及其算法[J]. 电子与信息学报, 2012, 34(5): 1142-1147.

Xu Jian-lou, Feng Xiang-chu, and Hao Yan. Improved TV-Stokes model and algorithm for image inpainting[J]. Journal of Electronics Information Technology, 2012, 34(5): 1142-1147.

Guillemot C and Le Meur O. Image inpainting: overview and recent advances[J]. IEEE Signal Processing Magazine, 2014, 31(1): 127-144.

Criminisi A, Prez P, and Toyama K. Region filling and object removal by exemplar-based image inpainting[J]. IEEE Transactions on Image Processing, 2004, 13(9): 1200-1212.

Sun Jian, Yuan Lu, Jia Jia-ya, et al.. Image completion with structure propagation[J]. ACM Transactions on Graphic, 2005, 24(3): 861-868.

Wexler Y, Shechtman E, and Irani M. Space-time completion of video[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2007, 29(3): 463-476.

白键, 冯象初, 王旭东. 图像分解的多尺度变分模型[J]. 电子与信息学报, 2013, 35(5): 1190-1195.

Bai Jian, Feng Xiang-chu, and Wang Xu-dong. A multiscale variational model for image decomposition[J]. Journal of Electronics Information Technology, 2013, 35(5): 1190-1195.

Barnes C, Goldman D B, Shechtman E, et al.. The PatchMatch randomized matching algorithm for image manipulation[J]. Communications of the ACM, 2011, 54(11): 103-110.

Zheng E, Dunn E, Jojic V, et al.. PatchMatch based joint view selection and depthmap estimation[C]. Computer Vision and Pattern Recognition (CVPR), Ohio, 2014: 1510-1517.

Cozzolino D, Poggi G, and Verdoliva L. Copy-move forgery detection based on PatchMatch[C]. IEEE International Conference on Image Processing (ICIP), Paris, 2014: 5312-5316.

He Kai-ming and Sun Jian. Computing nearest-neighbor fields via Propagation-Assisted KD-Trees[C]. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Judith, 2012: 111-118.

Gieseke F, Heinermann J, Oancea C, et al.. Buffer k-d trees: processing massive nearest neighbor queries on GPUs[C]. Proceedings of the 31st International Conference on Machine Learning, Beijing, 2014: 172-180.

Zhang H, Berg A C, Maire M, et al.. SVM-KNN: discriminative nearest neighbor classification for visual category recognition[C]. IEEE Computer Society Conference on Computer Vision and Pattern Recognition, New York, 2006: 2126-2136.

Ran Ling-qiang and Meng Xiang-xu. Fast seam carving using Gaussian pyramid[C]. Intelligent Human-Machine Systems and Cybernetics (IHMSC), Hangzhou, 2014: 59-63.

Ren Shuai, Lei Jing-xiang, Zhang Tao, et al.. Research of high performance information hiding scheme based on Gaussian pyramid and CARDBAL2 multi-wavelet for secret communication[J]. International Journal of Applied Mathematics and Statistics, 2014, 52(6): 234-251.

Yang Q, Tan K H, and Ahuja N. Real-time O(1) bilateral filtering[C]. IEEE Conference on Computer Vision and Pattern Recognition, Florida, 2009: 557-564.

施引文献

资源附件(0)

访问统计