Evaluating Information Loss in Digital Image Compression Techniques

Abstract

Uncompressed multimedia (graphics, audio and video) data requires significant storage capacity and transmission bandwidth. Despite rapid progress in mass-storage density, processor speeds, and performance of digital communication systems, demand for data storage capacity and data-transmission bandwidth outperform the capabilities of available multimedia technologies. The recent developments of multimedia-based applications have contributed not only for efficient ways of encoding signals but also in compression of signals. Therefore, the theory of data compression becomes more and more significant for reducing the data redundancy to save more hardware space and transmission bandwidth. In computer science and more specifically in information theory, data compression is the process of encoding information using fewer bits or other information-bearing units than an unencoded representation. Data compression is useful because, it supports to reduce the consumption of expensive resources such as hard disk space or transmission bandwidth. Image compression is an application of data compression on digital images, as it reduces the computational time and consequently the cost of image storage and transmission. The fundamental concept about image compression is to remove redundant and unimportant data, and at the same time to keep the compressed image with acceptable quality. In this paper, a comparison of information losses caused by three different image compression techniques were performed. The information loss was measured using the entropy of the image after the compression. The first compression technique is the Joint Photographic Experts Group' or (JPEG) which is based on the block based Discrete Cosine Transform (DCT) method. The second technique, which is based on the wavelet transform, where the testing was carried out using three types of wavelet functions namely: Haar, Morlet and Meyer. The third and the last method which is called seam carving which is an image resizing algorithm by establishing a number of seams in an image and by removing the seams, the image can be downscaled. The experimental results verify the ability of the different compression techniques based on the post compressed image entropy value. MatLabTM was used for the comparisons.