Lossless Image Compression Based On Layered Coding With Sequency Interpolation, And Its Implementation On DSP

The purpose of lossless image compression is to reduce the amount of data used to represent the image without any information loss, so the original image can be recovered correctly when decoding. It's mainly applied in some applications where the images are subjected to further processing, repeated compression/decompression, high cost of the image acquiring, the desired quality of the rendered image is yet unknown. Such as medical images, remote sensing image, image archiving, high-precision image analysis. Efficient, real-time lossless image compression has always been a goal pursued by the people. Therefore, developing a real time compression method based on DSPs are of significant practical and commercial interest.This paper primarily studies the lossless image compression based on layered coding technology with sequency interpolation, subsampling, and its implementation on the DSP development platform. First of all, superimposing digital filter and how to construct a two-dimensional IDCT low-pass limited superimposing digital filter are introduced in this paper. Secondly how to subsample the image and how to interpolation reconstruct the sampling data are discussed. In addition, the realization of lossless compression coding process is introduced.In the following, how to interpolate and reconstruct the image is analyzed, the IDCT low-pass limited filter is chosen as an interpolation template in this paper. From the respects of compression rate and the complexity of the operation, it identifies the optimal amount of layers which images are sampled and the coding technology for respective residual difference data and the top level data. Through the experiment of the test image, the optimal subsampling layers is determined as six and self-adaptive arithmetic coding is used to handle the residual difference data and top layer data. In addition, the realization process of the adaptive arithmetic coding is introduced in detail. The lossless compression for images by this arithmetic is realized with C program in the console of VC6.0. The code rates of lossless compression for classic test images are presented. At the same time, when use layered transmission, the relationship between the number of residual difference layers and the reconstructed image quality is provided.Finally this algorithm is realized in the DSP development platform. In this system, after the TI's TMS320DM642DSP is selected, the program is migrated into DSP and the image capture driver module and display module are also designed, making the compressing system based on DSP realized. To improve procedures for the operation of speed, the program is optimized combined with the hardware features of the C64xDSP.