Research Of Image Compression Coding Based On Regions Of Interest(ROI)

Because in recent years multimedia and the networking developed speedily, the image processing is used for broader domain,therefore more new request had been proposed to the image processing.Image compression code technology based on Regions of Interest arises under such background.Image compression code technology based on Regions of Interest,means to compress interesting regions of an image lossless or near-lossless, to compress background regions of an image a little loss.Like this can enable the reconstructed image have high signal-to-noise ratio(PSNR) and high compression ratio,solves the contradiction between the compression ratio and the image quality well.The article first introduces the research background and the domestic and foreign research present situation of image compression code technology based on Regions of Interest at present,introduced with emphasis the classic image code method on Regions of Interest.Next introduces the extraction of Regions of Interest by edge examination, make the detailed introduction to the use of edge operator and the wavelet method to the edge examination.The emphases of this article are the near-lossless image compression on Regions of Interest and a litter loss compression on Regions of Interest.In the foundation of promotion wavelet, realize Near-lossless image compression on Regions of Interest with lossless compression coding and embedded zerotree wavelets(EZW) coding. The experimental results indicates that image compression ratio decreased and the signal-to-noise ratio incresed as the size of ROI increasing.When the ROI size accounts for the entire picture size below 10%,this method usability is better,can obtain the greater compression ratio and the better composition picture.Moreover,this article realizes a little loss image compression coding based on Regions of Interest (ROI) by using the set partitioning in hierarchical trees algorithm (SPIHT) and the embedded block coding with optimized truncation (EBCOT). The experiment indicated that, under the same code rate,the Regions of Interest can be reconstructed by high quality than the background region. Finally ,designs the system of Regions of Interest (ROI) image compression coding based on TI TMS320C5402.According to the TMS320C5402 hardware characteristics as well as the compression requests, carries on the analysis to integer 5/3 wavelet, and from reduces the memory and reduces the repetition scanning number of times two aspects proposed the improvement program to SPIHT,enhances the coding efficiency.