Research On Efficient Image Coding And Post-processing Algorithm

In recent years,with the rapid development of Internet and the popularization of multimedia terminals,Image-based multimedia content appears in various fields such as media,medical care,surveying and entertainment.In order to solve the problem of image transmission and storage,researchers have already proposed many image coding and postprocessing algorithms.However,the current existing image coding and post-processing algorithms have low coding efficiency,and there is still room for improvement.In 1991,the International Organization for Standardization(ISO)and the International Telecommunication Union(ITU)developed the first JPEG(Joint Photographic Experts Group)standard.JPEG compression is widely used as a universal standard because it can be applied to any kind of image with low computational complexity,excellent compression ratio and small image loss.All of the current mainstream compression algorithms are evolved from this classic algorithm.Although JPEG has many advantages,its quality under high compression ratio is not good enough.There is a serious quality loss,affecting people's perception.The main steps of JPEG coding are transform,quantization,entropy coding,etc.In this thesis,we optimize one or more steps of JPEG to gain a higher coding efficiency,and three algorithms are proposed: image down-sampling algorithm based on rate-distortion optimization,quantization algorithm based on error compensation and compression constrained deblocking algorithm.The image down-sampling algorithm based on rate-distortion optimization is an iterative algorithm.The optimization process makes the decoded image with high quality based on the prior information of the original image.In addition,this algorithm can not only make the image in the transform domain coefficients sparse,but also set part of the transform coefficient to 0,so that we can directly discard these coefficients in the encoding process to implement downsampling in transform domain.It helps us to achieve rate-saving purpose.The quantization algorithm based on error compensation aims to reduce the distortion of RGB color images.The algorithm firstly converts the color image into YCbCr space,and compensates the Cb component with quantization error of Cr component,compensates the Y component with quantization error of Cr and Cb components.In this way,the mean square error of pixels in RGB space can be reduced and the quality of color image encoding can be improved.Quantization algorithm based on error compensation has good universality,which is suitable for most of the encoding processes.The compression constrained deblocking algorithm is an image filtering process.It not only smoothes the boundary but also ensures the coefficients of the transform domain unchanged as far as possible,which means that the block effect can be effectively removed and the entire image will not be distorted again by the filtering.Thus,the filtered image has a good objective and subjective quality.