Parametric Slant-Haar Transforms With Applications In Color Image Compression

A new class of non-sinusoidal unitary transforms called parametric slant-Haar transforms (PSHT) was recently proposed. It is an improvement of traditional slant-Haar transforms that has led to great variation and flexibility. In this paper, the recursive generation of the parametric slant-Haar transform matrix that corresponding to four classic orders of Walsh transform is presented respectively, and their fast algorithm is designed by the factorization of transform matrix.In addition, we propose a new color image compression and coding scheme based on JPEG base-line color system. It includes the following improvements:(1)utilizing three-dimension transform instead of two-dimension transform.(2)replacing the nearly optimum DCT by PSHT.(3)altering the block size from 8×8 to more effective 16×16.Both theoretic analysis and experiment results demonstrate that the PSHT exhibits low computational complexity as well as good performance, and the proposed coding scheme base on 3D-PSHT has a good application prospect.