The Software Emulation And Hardware Implemetation Of Compress / De-compress Method Base On CFA

Consumer electronics field comes into high definition video era now. The information carried by HD video is much huger than standard definition video. The huge information is not suitable for storage, transaction and carriage. So many IC design companies spend much effort on data compression and decompression just because of the size and power dissipation has the same meaning as money for IC designs. By common way used in compression(MPEG-2, MPEG-4, h. 264, VC-1), the data is compressed according to the characteristics of the image after it were collected. The huge calculation and large storage space in front compression module is inevitable. The method discussed here for compression and decompression is different from the usual way. We expect to find a way to compress the data from the source in order to have a sharply decline of calculation from the video source. All the video-system can benefit from the small calculation.For compression, we will refer to Bayer Pattern when we compress the video source. When we do software emulation with Matlab, we use Matlab embeded function to read image source and separate the RGB source, and then do compression. After compression, the size is compressed to 1/3 of the source size.For decompression, we will do decompression with the usual method which is already commonly used in industry. These methods include bilinear interpolation, constant hue based interpolation, color difference interpolation, and bilinear interpolation with laplacian second order correction. We will use matlab to recover the RGB data based on compression method, and write out image file in ‘bmp’ format, and calculate the PSNR value of the image. According to the emulation result, we found a better way to do compression which named bilinear interpolation with edge directed based on YCbCr color space.We choose the bilinear interpolation with edge directed based on YCbCr color space to do RTL implementation. The interpolation method costs small calculation, refers to less pixel information, and has better decompression quality. This discourse introduces the architecture definition, the implementation with verilog HDL, and the verification method. All of these have proven the correctness of the decompression method.This project has proven that the bilinear interpolation with edge directed has a better performance in image decompression.