Image Compression Based on Hybrid Wavelet Transform Generated using Orthogonal Component Transforms of Different Sizes

In this paper, image compression using hybrid wavelet transform is proposed. Hybrid wavelet transform matrix is generated using two component transform matrices. One component transform matrix contributes to global properties whereas second one contributes to local properties of an image. Different sizes of component transform matrix can be used to generate hybrid transform matrix so that its size is same as image size.Different colour images of size 256x256 are used for experimentation. Proposed hybrid wavelet transform is applied on red, green and blue planes of image separately. Then in each plane transformed coefficients are sorted in descending order of their energy and lowest energy coefficients are eliminated. Root mean square error between original image and reconstructed image is calculated to check the performance at different compression ratios. By varying the size of pair of component transform matrices, hybrid transform matrix is constructed and results are observed.Also by changing the component matrix which contributes to local properties of image and with size variation, results are observed and compared. It has been observed that if more focus is given on local features, then better results are given by that Hybrid Wavelet Transform. Focusing on local features can be done by selecting larger size of orthogonal component transform that contributes to local properties.