Performance Analysis of Vector Quantization Based Lossy Image Compression

This paper presents a Lossy image compression technique which is combination of discrete wavelet transform (DWT), Thresholding, Vector Quantization (VQ) and Huffman coding. Proposed method is as follows, First, DWT is performed on the original image then Globle Thresholding technique is applied and resulting coefficients are then vector quantized. VQ indices are Huffman coded to increase the compression ratio. There are many lossy techniques exists for image compression in digital domain, among this wavelet transformation based image compression by using vector quantization (VQ) provides good picture quality and better image compression ratio compared to all other techniques. Vector quantization (VQ) has the potential to greatly reduce the amount of information required for an image because it compresses in vectors which provides better efficiency than compressing in scalars. The objective is to generate the standard codebook by using some standard training set which is capable of successfully coding images outside of the training set. Vector quantization (VQ) based coded images then encoded for transmission by using Huffman encoding. A SSIM (Structural similarity index measurement) is used to check the image quality result are measured in terms of Compression Ratio (CR), peak signal to noise ratio (PSNR) of the reconstructed image and Root Mean square error (RMSE) The proposed Lossy image compression techniques gives higher Compression Ratio with better image quality in terms of PSNR, SSIM, compared to other DWT and VQ based image compression techniques. The proposed method of image compression is useful for various applications criminal investigation, medical imaging, etc