Image Compression using Fusion of Hybrid Wavelet Transform and Vector Quantization

This paper proposes novel lossy image compression technique using hybrid wavelet transform and vector quantization. First hybrid wavelet transform consisting of two different component transforms is generated and applied on color images. Discrete Kekre transform (DKT) and Discrete Cosine transform (DCT) play role of base and local transform respectively in hybrid wavelet transform. In transform domain 3.125% data is retained by making low energy coefficients zero and image is reconstructed. Vector quantization (VQ) algorithm is applied on these reconstructed images. In vector quantization only indices of code vectors are sent which gives more compression of hybrid wavelet transformed image. Three different vector quantization algorithms like Linde-Buzo-Gray (LBG), Kekre’s Proportionate Error (KPE) and Kekre’s Error Vector Rotation (KEVR) are applied on reconstructed images and their performance is compared. Error between original image and reconstructed image obtained after vector quantization is calculated and reconstructed image quality is observed for each VQ algorithm. In proposed method KPE algorithm shows better image quality than traditional LBG algorithm. Results of KPE are followed by KEVR algorithm. Combination of hybrid wavelet transform and VQ helps to achieve higher compression ratio up to 64 giving better quality of reconstructed image than obtained in DKT-DCT hybrid wavelet transform. KeywordsImage Compression;, Kekre Transform; KPE; KEVR; Vector Quantization. African Journal of Computing & ICT Reference Format: H.B. Kekre, T. Sarode & P. Natu (2014). Image Compression using Fusion of Hybrid Wavelet Transform and Vector Quantization. Afr J. of Comp & ICTs. Vol 7, No. 5. Pp 85-94.