Research Of High Definition Image Compression Technology Based On JPEG XR

With the rapid development and popularization of multimedia technology and wireless communication technology, people are eager to acquire more images. At present, the size of images is always big, so there is a challenge for the limit of channel transmission bandwidth and storage space. Image compression is an effective way to solve this contradiction. A valid and common method to solve this problem is image compression. Now there are three static image compression international standards, the Joint Photographic Experts Group released the newest static image compression standard-JPEG XR in 2007. Its core transform is Lapped Biorthogonal Transform(LBT), compared to JPEG that is most widely used, JPEG XR over the deficit of block artifacts at low bit rate and improve the quality of image. JPEG XR can almost get the same quality of image that compressed with JPEG2000, but the source consumption is very close to JPEG. JPEG XR supports coding of the region of interest, thumbnail extraction. Based on the JPEG XR compression standard, we mainly do the research from the following several aspects:1. Aiming at manually selecting the shortage of the region of interest in the region of interest coding, this paper puts forward to the image compression algorithm based on automatic extracted ROI. In this paper, we use edge detection operator and dilation and erosion operation to extract the region of interest from the edge, according the edge of the block transform coefficients belonging to the proportion of the ROI, we can determine whether the block belongs to the edge ROI block. We can further determine for the block that contains both ROI information and BG information whether should be allocated more bit rate. This can be a very good weaken the BG information in the edge of the ROI.2. Aiming at the problem of the large-scale face image recognition system in network transmission and storage, this paper evaluates the effect of JPEG、JPEG2000 and JPEG XR on selected face recognition algorithms. By using data compression techniques, it is possible to remove some of the redundant information contained in images, requiring less storage space and less time to transmit. The face images are high compression ratio in the face database, and tested in several recognition algorithms. The experimental result shows that the face images are tested the recognition rate at high compression. In a number of recognition algorithm selected, the SIFT is better. The test data show that three kinds of compression algorithm have effect to identify the compressed rate. Through comprehensive analysis and comparison, we finally conclude that JPEG XR is low in complexity and performance on the recognition rate.3. Aiming at the block effect that is caused by JPEG XR under high compression ratio, this paper put forward a method of weakening the block effect that uses the 1D and 2D dual-tree complex wavelet transform of post-processing technology. We use the paper of the image processing technology with block effects that were deblocking processing by performing LBT transformation to reconstructed image obtained and setting the appropriate threshold. The experimental results show that this algorithm can effectively weaken due to the high compression ratio of the block effect, and the main information of the image is preserved and improves the subjective visual quality.