A Study Of And The Exploration For The Key Algorithm Of JPEG2000Standard

JPEG2000is the new international standard for image compression, which notonly has superior compression performance but also offers some salient features, suchas progressive transmission by pixel accuracy and resolution, random access andprocessing of stream, therefore, JPEG2000standard will be effective in wideapplication area such as Internet, remotely sensed satellite imagery and printing.However, these excellent features were achieved at the cost of drawback that algorithmcomplexity and computation are about30times higher than that of JPEG. Therefore,in-depth research is devoted to the key algorithm of JPEG2000, and some solutions areproposed to improve the efficiency of compression or the quality of the decompressionimage, to significantly decrease the algorithm complexity and computation and improvethe real-time of encode system at almost the same quality of the decompression image.On the other hand, based on these research and exploration, a new image compressionalgorithm is proposed, in order to facilitate practical application of engineering.Discrete Wavelet Transform is the key transform algorithm of JPEG2000compression standard, and Embedded Block Coding with Optimized Truncation is keycoding algorithm of that. In this paper, in-depth research is firstly devoted to the keyalgorithm of JPEG2000, then the data stream of the encoding process of JPEG2000isanalyzed and explored, and also some improved methods are proposed, such as a studyof interpolation algorithm of wavelet sub-band, making advantage of the position similitudebetween the near wavelet sub-band coefficients to do some interpolation extrapolation; a study ofthe encoded bit-plane scanning model, correcting the coding model to improve he data codingefficiency; a study of bit-plane truncation optimization algorithm, a new rate controlalgorithm is proposed to be "on demand code"; a study of image post-processing algorithmof model-based edge, reconstructing the degraded edges with different-degree compensation.The research feasibility and expected results of these algorithms are of consider.