Research On CFA Data Image Compression Based On JPEG-LS Algorithm

JPEG-LS is the algorithm of the new ISO/ITU standard for lossless and near-lossless compression of continuous-tone images. It possesses High-Fidelity and Low-Complexity and is applied to Digital Camera,Network Transmission,Wireless Communication and Medical Image, etc. CFA data image came from CCD sensor is un-processing. Compared to common image, every pixel of the CFA data image has a color value only. To the special data, compressing and storing directly have the advantages of lower work and lower redundance, but can offer more information. So it has important meaning to channel transmission of data and next processing. At present, lossless-compressing of CFA data image is starting. There is no algorithm which has high C/R. The compression ratio is about 1.6. Based on research of JPEG-LS algorithm, this paper has analyzed the statistical character of CFA data image. The C/R has been enhanced to over 1.9 after construction separation and conversion. The main job of this paper as follows:(1) Research on statistical character of image. Chapter 2 gives mathematics describe of image, analyzes determination and statistic of image data, and gives model of image. In this chapter, the nature image is almost seemed as Markov process. In this process, the pdf(probability density function) of a pixel value is determined by some neighborhood pixels. To CFA image of single CCD, we separate the image into 3 color plates, and regard each as a gray image. To each gray image, we do an autoregressive model, and confirm a fit orders to process "context" predicting.(2) Research on the core algorithm LOCO-I of JPEG-LS. We study 6 parts of the algorithm, including Prediction,Context Modeling,Adaptive Correction,Parameter k Estimation,Re-mapping,Adapted Limited-length Golomb Coding. Specially, we deduce the Parameter k of Golomb code in theory. At last, we give an example of JPEG-LS code.(3) Compression experiment on CFA data image. In this paper, we give 4 methods. First, compressing directly to CFA data image; second, compressing based on construction separation and conversion. This means comprise no filtering,mean low-pass filtering,Gauss low-pass filtering; Third, compressing based on modified JPEG-LS algorithm, including modified predictor and parallel processing; Fourthly, compressing based on some standard compression algorithms, including JPEG2000, L-JPEG,PNG. At last, we have compared the efficiency of JPEG-LS to the others.