Researches On Image Compression Algorithm And Its Hardware Implementation For Remote Sensing Images

The imaging system for satellite remote sensing was widely used in many fields such as mapping, intelligence and reconnaissance, resource exploration, environmental monitoring and lunar exploration, with which plenty of information on earth's topography, geomorphic as well as geological structure could be obtained. With the increasing development of the image sensing technology, the resolution of remote sensing image became larger and larger, and the amount of image data required to download increased dramatically. It was consequently necessary to compress remote sensing images efficiently on the satellite because the bandwidth of satellite downlink channel was limited. Therefore, there was important theoretical significance and application value in the study of high-performance on-board image compression algorithm and its hardware implementation for satellite remote sensing images.The in-depth analysis and study of the core algorithm of JPEG2000 and SPIHT was presented in this paper. By optimizing the encoding structure and process of the original algorithm, novel structure of wavelet transform, JPEG2000 rate control and region of interest coding algorithms were proposed. Compared with the traditional methods, the proposed algorithms and structures had the advantages of high-speed data processing, low complexity computation and low memory usage, which was suitable to be applied by the satellite imaging system. Based on the new algorithms, a hardware system for remote sensing image compression was designed and implemented on FPGA. The main achievements of this paper can be summarized as follows.1. A novel rate control algorithm with its hardware structure for JPEG2000 was presented, which was prone to be implemented. After wavelet transform and quantification on original image, a prediction module is introduced to estimate the entropy of EBCOT code block within available bit-plane, then using the proportion between the estimate entropy of each code block and the summation estimate entropy of all code blocks to allocate the rate for each code block. The proposed algorithm can not only efficiently reduce the cost of computation and memory usage but also have the advantages of high quality image reconstruction and simple hardware structure.2. A new algorithm for ROI coding was proposed based on rate pre-allocation. Code blocks of ROI were primarily encoded and its coding stream length was controlled to implement lossy to lossless ROI coding. Certain bytes were allocated to each block of background region(BG) according to its entropy value. In the light of allocated rate, T1 coding process on block of BG was able to be real-time terminated. As a result, the whole coding process was simplified.3. A new compression algorithm for interferential multi-spectral images was proposed, which was based on template matching algorithm and region of interest coding method. In order to eliminate the redundancy between the two successive frames, the template matching algorithm in the wavelet domain was adopted. With the method based on ROI coding of EBCOT, the image produced after template matching was coded. The proposed algorithm can improve the reconstructed image quality, enhance the efficiency of image coding and efficiently protect the multi-spectral information.4. A VLSI architecture of rate control algorithm for JPEG2000 standard was presented. By means of improving the method to estimate rate-distortion slope and optimizing the procedure of code stream truncation, the data processing speed of T2 encoder was effectively enhanced, the hardware structure complexity of that was reduced. Furthermore, layered stream output and ROI coding were implemented well.5. An improved wavelet transform algorithm and an improved SPIHT bit-plane coding algorithm were proposed. Based on them a hardware system for image compression was designed and implemented on FPGA. The data processing speed of improved structure of wavelet transform algorithm was increased significantly, the external memory usage of that was reduced. The anti-error performance of the improved SPIHT coding algorithm was enhanced by the way of segmentation coding in wavelet domain, the code stream organization of that was implemented efficiently with the use of rate control algorithm for JPEG2000 and ROI coding was functioned using the rate-distortion slope lifting method.