JPEG-XR Image Compression Coding Optimization Based On Coefficient Prediction

With the development of digital image technology and the popularization of multimedia applications,image processing and image transmission have become more and more important.People's requirements for image quality have gradually increased.For example,in the fields of telemedicine,aerospace,multimedia teaching,and video security,higher requirements have been placed on image transmission and storage.Therefore,new technologies and research works on image compression have brought new challenges.Almost all multimedia applications desire image compression technologies with higher compression ratio,lower computational cost and better visual quality,which are also the three fatal indicators for image compression.JPEG-XR(formerly known as HD Photo)is a continuous-tone still image compression algorithm.It is the latest still image compression standard released by the Joint Photographic Experts Group in 2007.It has high compression ratio and low calculation cost.It supports both lossy and lossless encoding of still images.Compared with the most widely used image compression standard JPEG JPEG-XR uses a more efficient compression algorithm to double the image compression capability,and JPEG-XR does not lose any information under the high quality compression.Compared with the image compression standard JPEG2000,JPEG-XR consumes less computational cost to obtain the same quality image after compression.In order to pursue higher image compression efficiency,this research is based on the JPEG-XR image compression standard,and made in-depth research from the following aspects:(1)This thesis propose,JPEG-XR-GCP,a gradient-based coefficient prediction(GCP)to replace the prediction in original JPEG-XR.It predicts the coefficients based on the estimating of local gradients,and predicts the value of coefficients adaptively by using its neighboring and second-order neighboring coefficients in gradients.(2)The scheme GCP improves the performance of standard JPEG-XR both in lossless and lossy compression modes.Compared with standard JPEG-XR,our proposal reduces the average bit rate from 0.25 to 1 in lossless compression,and improves the average PNSR of the compressed image from 0.3DB to 0.7DB in lossy compression.(3)This thesis conduct extensive experiments on several publicly available test databases to evaluate the performance of the proposed JPEG-XR-GCP.Experimental results demonstrate that JPEG-XR-GCP scheme outperforms the original JPEG-XR and the existing other improvements on it.