Performance Evaluation of Lossy Image Compression Techniques Based on the Image Profile

Abstract

Digital images have become part of our everyday life. It is important to store these images in an efficient manner in the given storage space and therefore digital image compression has become a large focus in the recent years. Image compression deals with the rebate of bits required to epitomize the image. It is also important to ensure that reduction in size of image is without degradation of image quality and loss of information. As a result of this, image compression techniques have been developed with lot of new algorithms and also with variations of the already existing ones. For a given application, the image compression algorithm is selected with an objective either to provide better compression or to get a good quality of reconstructed image. The profile of the image such as aspect ratio, color, pixel intensities, smooth regions, edges, shading, pattern, and texture is not considered while selecting an algorithm. In this paper, an attempt is made to provide the users a ready reckoner to select a compression technique based on the profile of the image. In this regard, three popular techniques such as set partitioning in hierarchical tree (SPIHT), compressive sensing (CS), and fractal coding (FC) in the family of lossy image compression are selected for analysis. In addition to this, the best choice of wavelet filter banks, measurement matrix, and reconstruction algorithm are also identified to provide improved performance. The performances of these algorithms are evaluated with the metrics such as compression ratio (CR), mean square error (MSE), and peak signal-to-noise ratio (PSNR).