350 rub
Journal Achievements of Modern Radioelectronics №3 for 2012 г.
Article in number:
Non-Local Principal Component Analysis Image Filtration Algorithm
Authors:
E.V. Sergeev, I.S. Mochalov, V.A. Volokhov, A.L. Priorov
Abstract:
In any image taken with the help of optoelectronic converters, there is one or another part of the additive noise. This work is devoted to modern methods of the noise removal, based on a non-local treatment. A Non-local processing is based on a generalization of Yaroslavsky on the case where each pixel of the digital image is described not only by its brightness value, but also the values of its neighbors. Natural property of this approach is to increase its efficiency while increasing the size of images. It is certainly a positive fact, since the resolution of modern camerasan is constantly growing. The article describes a new method of image filtering based on the multistep filter system. Processed data for each filter is the same noisy image. The result of the filtering in the previous stage is used only indirectly, as an estimation of the constructed new non-local filter. Nonlocal filter is constructed by finding for each block a set of similar blocks and assigning each of them a weight depends on the Euclidean distance between them. The first step of this algorithm is a hard threshold of coefficients of the undecimated wavelet transform (Donoho scheme). Result is then used for block matching to find a set of similar blocks that then processed in the basis of principal components (Karhunen-Loeve basis). Filtered image itself becomes the estimate for block matching for the Wiener filtering in the basis of principal components. Finally, the filtered image is fed to the post-processing algorithm that enhances the image details that were lost in the previous steps. Such scheme allows to achieve significant results in the processing of a wide class of noisy images
Pages: 80-88
References
  1. Chatterjee P., Milanfar P. Is denoising dead - // IEEE Trans. Image Processing. 2010. V. 19. № 4.
  2. Katkovnik V., Foi A., Egiazarian K., Dabov K. From local kernel to nonlocal multiple-model image denoising // Int. J. Computer Vision. 2010. V. 86. № 8.
  3. Nadaraya E. A. On estimating regression // Theory Probab. Applic. 1964. V. 9.
  4. Watson G. S. Smooth regression analysis // Sankhya Ser. 1964. V. 26.
  5. Fan J., Gijbels I. Local Polynomial modeling and its applications // Chapman and Hall. 1996.
  6. Yaroslavsky L. Digital picture processing - an introduction // Springer. 1985.
  7. Tomasi C., Manduchi R. Bilateral filtering for gray and color images // Proc. VI Int. Conf. Computer Vision. 1998.
  8. Saint-Marc P., Chen J. S., Medioni G. Adaptive Smoothing: A General Tool for Early Vision // IEEE Trans. Pattern Analysis and Machine Intelligence. 1991. V. 13. № 6.
  9. Comaniciu D., Meer P. Mean shift analysis and applications // IEEE Trans. Pattern Analysis and Machine Intelligence. 1999.
  10. Моисеев А. А., Кобелев В. Ю., Волохов В. А. Курвлет-преобразование в задаче подавления шума в изображениях // Цифровая обработка сигналов. 2008. № 1.
  11. Приоров А.Л., Волохов В.А., Мочалов И.С. Синтез двумерных неразделимых вейвлет-фильтров для субполосного разложения произвольной кратности // Радиотехника. 2010. № 1.
  12. Muresan D. D., Parks T. W. Adaptive principal components and image denoising // IEEE Int. Conf. Image Processing. 2003. V. 1.
  13. Zhang L., Dong W., Zhang D., Shi G. Two-stage image denoising by principal component analysis with local pixel grouping // Pattern Recognition. 2010. V. 43. № 8.
  14. Buades A., Coll B., Morel J. M. A review of image denoising algorithms, with a new one // SIAM Multiscale Modeling and Simulation. 2005. V. 4.
  15. Dabov K., Foi A., Katkovnik V., Egiazarian K. Image denoising by sparse 3D transform-domain collaborative filtering // IEEE Trans. Image Processing. 2007. V. 16. № 8.