A Combined Approach for Lossless Image Compression Technique using Curvelet Transform

Abstract — Image compression is an unavoidable research area which addresses the problem of reducing the amount of data required to represent a digital image for minimizing the memory requirement and system complexity. In the recent years, most of the efforts in the research of image compression focused on the development of lossy techniques. The key idea of our proposed scheme describes lossless compression using curvelet transform combined with error correcting BCH and modified arithmetic encoding technique. Most of the wavelet based approaches are well suited to point singularities have limitations with orientation selectivity and do not represent two dimensional singularities (e.g. smooth curves) effectively. Our proposed curvelet based approach exhibits good approximation properties for smooth 2D images. The BCH encoder converts the message of k bits in to a codeword of length n by adding three parity bits. The image can be divided into blocks of size 7 bits and entered to the BCH decoder which eliminates the parity bits. Thus the block of 7 bits will be reduced in to a block of size 4 bits and output will be in two folds. The first file contains the compressed image and the second contains the keys. The simulation results show that our proposed compression scheme gives more than 50 % memory saving at peak signal to noise ratio (PSNR) 45 dB with 0.5 bit per pixel (BPP)