Image Compression using Wavelet Transforms of DCT, DST, Hartley and Real-DFT with Variation in Size of Component Transforms

This paper presents simple wavelet based image compression method where wavelet transform is generated from orthogonal component transforms. Wavelets of DCT, DST and Real-DFT and Discrete Hartley Transform (DHT) are generated. Each generated wavelet transform is applied separately on R, G, and B plane of 256x256x3 colour image. Performance of these wavelet transforms is evaluated using two parameters. One is compression ratio and other is ratedistortion graph. Size of component transforms in their respective wavelet transform is varied and results are compared. From results it is observed that DCT and Real-DFT wavelet gives better compression using component transform size of m=8 and n=32. DST wavelets show considerable blocking effect. Blocking effect is more intense if local component transform size (i.e. ‘n’) is larger. These wavelet transforms are compared in terms of bit rate as a function of distortion. It shows that DCT wavelet gives best performance among all four. Real-DFT ranks second followed by Hartley wavelet and then Discrete Sine wavelet transform.