A Still Image Compression Scheme with Joint Probability Based Scanning of a Bit Plane Using Golomb-Rice Code

In this paper, a modified JPEG2000 still image compression system has been proposed. A three level decomposition of Daubechies 9/7 Discrete Wavelet Transformation has been first applied to the entire input image. Then, scalar quantization is used to decrease and round off the transformed coefficients. The quantized coefficients are then subjected to bit modeling in each bit plane. A joint probability statistical model based significance selection has been proposed to select the significant bit for entropy coding with two scan coding technique. In this proposed work, after selecting all the significant bits in a particular bit plane, a geometrically distributed set of context is modeled and subjected to encode with Golomb-Rice coding to give compressed data. The decompression is effected with a simple, respective inverse operation. The proposed system has been experimented with standard benchmark images and the standard performance measures, Compression Ratio and Peak-Signal to Noise Ratio are used to evaluate the result.