With the rapid development of remote sensing, more and more high-resolution satellite image is used in national defence and economy. Considering the bandwidth of transmission and the power of satellites, lots of datum of the satellite image must be compressed to meet the demand of the high-speed transmission and storage. Satellite image compression technology differs from normal image compression technology, because it is more restricted on hardware condition, and requires further real-time characteristic. Therefore, we need to develop a real time compression method which is suit for the hardware implementation.In 2000, JPEG2000 was approved as a new still image compression standard by a collaborative effort between the international Organization for Standardization (ISO) and International Electrotechnical Commission (IEC). JPEG2000 forms a whole suit of image compression standard by effectively integrating the two excellent fruits, wavelet transform theory and efficient data structure method: embedded block coding with optimized truncation (EBOCT), which were developed from these latest ten years digital signal process domain.As for satellite image, JPEG2000 algorithm based on discrete wavelet transform (DWT) is a good idea. But it must be pointed out that the implementation of the two dimensional discrete wavelet transform is of high computational complexity and needs much memory which is impossible in some particular conditions, such as satellite and portable devices. Because of this, the paper brings forward an improved method based on JPEG2000, which is suit for the fixed point implementation of DSP.The follow are the major study result of the paper.(1) We have studied the fast computation, low-memory realization and fixed-point implementation of the two dimensional discrete wavelet transform. It includes: an improved method based on spatial combinative lifting algorithm, which effectively reduces the number of multiplications; low-memory realization based on line; the improvement of the arithmetic shift; and the change of lifting parameters in the IDWT.(2) Through the isolation of signal coding and modification to context model, three coding passes are changed to two coding passes in the bit-plane scanning, as a result of the improvement of the algorithm in image coding.(3) Using the features of DSP, the modified algorithm of image compression based on JPEG2000 is optimized to enhance the capability of the parallel processing .The experiment of standard image compression test indicates that the performance of the algorithm on JPEG2000. is superior to the standard: the former requires less memory than the latter on the same quality of the picture. Thus, it is more competent in the fixed point implementation of DSP. |