Title :
Is the SVD-based low-rank approximation optimal?
Author_Institution :
Dept. of Electr. & Comput. Eng., Victoria Univ., BC, Canada
Abstract :
It is well known that the singular value decomposition (SVD) offers an optimal approximation in F-norm for a given image. In this paper, we show that using a 2D-to-4D mapping in conjunction with optimal 4-D low rank approximation, improved compression rate can be achieved compared to that obtained from the conventional (2-D) SVD approach
Keywords :
data compression; image coding; singular value decomposition; 2D-to-4D mapping; SVD-based low-rank approximation; compression rate improvement; image compression; optimal approximation; singular value decomposition; Bit rate; Data compression; Digital signal processing; Discrete Fourier transforms; Discrete cosine transforms; Discrete transforms; Image coding; Pixel; Signal representations; Singular value decomposition; Two dimensional displays;
Conference_Titel :
Circuits and Systems, 1995., Proceedings., Proceedings of the 38th Midwest Symposium on
Conference_Location :
Rio de Janeiro
Print_ISBN :
0-7803-2972-4
DOI :
10.1109/MWSCAS.1995.504478
