Singular Value Decomposition on Reducing Required Storage Space of Video Representation

Abstract

: The needs of data compression as in video compression emerge due to limitation of storage and communication bandwidth. Video is a sequence of digital images, where the digital images can be represented as matrices. One of the matrix factorization that is possible to be implemented on any matrix is Singular Value Decomposition (SVD). On low rank SVD, we choose a smaller value of rank of the new approximation image to reduce its necessary storage capacity. When we save the images in the storage by saving the new representation from SVD results other than the original representation, theoretically with smaller rank of the matrix representation will results in smaller size of file. The experiment result on this research gives a compression ratio 2.6182 on a sample frame of the video data by choosing text file format to save the necessary data.