Embedded Image Coding Algorithm Based On Block Dct Transform

Digital image compression is considered as one of the hottest research topics in information processing technology. Discrete Cosine Transform (DCT), due to its good relevance and energy-compression characteristics, has become the core technology of most current image coding standards. However, in the DCT transform coding, the entire image coded information can be reconstructed after full decoding, and image compression bit stream does not have progressive features. Therefore, it cannot satisfy the scalable bit-stream requirements of modern wireless communications. In this study, according to the non-progressive characteristics of stream problem in DCT transform coding, a block-based DCT embedded coding algorithm is proposed blocking artifacts caused by blocking were studied. The main contents of the study are as follows:1. After studying the basic knowledge of image coding, the traditional DCT transform coding algorithm, the classical EZW coding algorithm, and bit-plane coding, the author concludes the advantages and shortcomings of DCT transform coding, EZW, and bit-plane coding in order to put forward a new image compression coding algorithm based on this theoretical framework.2. A embedded coding algorithm is proposed based on block-based DCT (Block-DCT or BDCT). The algorithm takes full advantage of DCT transform coefficients of the natural tower structure, and uses symbols in place of numerical code to reduce the coding capacity. In addition, successive approximation quantization strategy is used, and the threshold value of the DCT transform coefficients is set for the detection importance. Transform coefficients are encoded according to the importance of image information from most important to least important The results show that the algorithm used in this thesis significantly improves the human visual quality of the reconstructed images and objective quality with a higher peak signal to noise ratio. The algorithm also achieves the image's embeddedness in the DCT domain. 3. A mean filter deblocking method is proposed in space domain. The border of the image pixels to be processed is included in the template as much as possible by setting the template. Variance theory is used to analyze the correlation between processed pixel and the surrounding pixels. Then according to the strength of correlation, the treatment is determined to deal with pixel intensity of the filter. The continuity of the border is restored and it improves the quality of the image problem caused by blocking artifacts on low bit rate in BDCT algorithm. Results show that this method improved the subjective visual quality and also enhanced the PSNR.