Summary

International Technical Conference on Circuits/Systems, Computers and Communications

2016

Session Number:M2-5

Session:

Number:M2-5-1

Fast Image Compression Using Enhanced Singular Value Decomposition

Harris Kristanto Husien,  Wannida Sae-Tang ,  

pp.199-202

Publication Date:2016/7/10

Online ISSN:2188-5079

DOI:10.34385/proc.61.M2-5-1

PDF download (1.6MB)

Summary:
This paper proposes a new method called as expected K-dimension (EKD) for dealing with SVD image compression. The hardest thing is to find the best K-dimension or the redundancy level. The reason is that if the K-dimension that being chosen is too high, it means there is redundant data (more information can be cut), while if the K-dimension is set too low, it means some important information can be missing. The proposed method finds the minimum gap between values in S matrix and then utilizing the remaining values to determine the K-dimension. The results show that the proposed EKD method gives the result as good as the conventional SVD in terms of compression ratio and PSNR, but it reduces the processing time and the complexity drastically.