IMPLEMENTATION AND COMPARISON OF IMAGE COMPRESSION TECHNIQUES

Abstract

In this thesis, a comparative study between transforms used for the compression of still images i.e. Discrete cosine transform (DCT), Karhunen Loeve Transform (KLT) and Discrete wavelet transform (DWT) is presented. MATLAB software is used to generate image compression results for algorithms using DCT, KLT and DWT. The Karhunen Loeve Transform (KLT) is optimal transform , as it packs most of the energy into a fewer number of frequency domain elements and it completely de-correlates its elements but it is theoretical concept & practically very difficult to implement as its basis functions are data dependent. In terms of complexity involved, DCT achieves the fastest computational performance due to fast algorithms. Wavelet Transform has superior performance than other transforms in terms of peak signal to noise ratio & image quality. DWT provides higher compression ratio due to the wavelet transform multi-resolution capability which allows it to analyze signal at various scale and resolutions. 4