Article

Article title IMAGE COMPRESSION BY USING TENSOR APPROXIMATION
Authors M.K. Tchobanou, D.V. Makarov
Section SECTION III. MODELLING OF COMPLEX SYSTEMS
Month, Year 02, 2013 @en
Index UDC
DOI
Abstract The paper considers the problem of image compression while images are supposed to be the multidimensional signals. There is introduced tensor approximation method, developed for high dimension data compression and enabling faster computing. The results of the two methods of Tensor-Train Decomposition (TT) and Wavelet Tensor-Train (WTT) for image compression are presented. It was found that the method WTT gives greater compression than the TT. WTT comparison with the popular image compression algorithms (JPEG and JPEG2000) shown that the use of low rank filters WTT loses, and if the filters of large rank exceeds JPEG and JPEG2000.

Download PDF

Keywords Multidimensional signal; tensor approximation; lossy compression; reduction of the signal space dimension; approximation chain tensors; signal functionally-specific filters.
References 1. Orfanidis, S. SVD, PCA, KLT, CCA, and All That // Optimum Signal Processing, Rutgers University. – 2007.
2. Dasgupta S., Gupta A. An elementary proof of a theorem of Johnson and Lindenstrauss // Random Structures and Algorithms. – 2003. – Vol. 22. – P. 60-65.
3. Oseledets, I.V. Tensor-train decomposition // SIAM J. Sci.Comput. – 2011. – Vol. 33, № 5. – P. 2295.2317.
4. Oseledets, I.V. Approximation of matrices with logarithmic number of parameters // Doklady Math. – 2009. – Vol. 428, № 1. – P. 23-24
5. Wallace G.K. The JPEG Still Picture. Compression Standard // Communication of the ACM. – 1991. – Vol. 34. – № 4.
6. Чобану М. Многомерные многоскоростные системы обработки сигналов. – М.: Техносфера, 2009. – 480 c.

Comments are closed.