Research On Remote Sensing Image Processing Combining High Fidelity Compression And Resolution Enhancement

The rapid increasing data of the remote sensing image (RSI) brings great challenges to storage, transmission and application. Therefore, it's becoming more and more important to compress RSI efficiently. Most conventional compression methods are blindly pursuing the high subjective quality or objective quality, neglecting the special applications for RSI. In order to get high SNR, these methods tend to discard high-frequency information,and as the results the resolution of reconstructed image is declined. It's very negative for RSI which has abundant high-frequency information and special applications in military reconnaissance. The combined processing technique, both considering the quality of reconstructed image and special applications, can effectively avoid shortcomings of conventional compression, and be more suitable for RSI compression. Therefore, the research on remote sensing image processing combining high fidelity compression and resolution enhancement has great theoretical significance and application value.Being a new international still image compression standard, JPEG2000 provides a new solution for RSI compression. Its excellent rate-distortion performance insures the quality of reconstructed image on the whole. However, as a common standard, JPEG2000 has some limitations for RSI compression, and it is necessary to improve it according to the characteristics and special applications of RSI. Therefore, this dissertation, based on JPEG2000 standard, studies the combined processing technique which associates high fidelity image compression with resolution enhancement, then realizes RSI compression in the hardware system.Firstly, the fundamentals and coding characteristics of JPEG2000 are thoroughly reviewed, then compares the JPEG2000 with conventional compression methods in effect of RSI compression. In order to analyze the decline in resolution during the process of compression, this dissertation gives theoretical analysis and takes same experiments. From both theoretical and experimental analysis, JPEG2000 standard tends to discard high-frequency information,and as the results, the resolution of reconstructed image is declined.Secondly, JPEG2000 tends to discard high-frequency information to get high quality, but the resolution of reconstructed image is declined. In order to solve this problem, the research on remote sensing image processing combining high fidelity compression and resolution enhancement is proposed. This combined processing algorithm, taking full advantage of multi-resolution decomposition and time-frequency analysis of wavelet transform, detects important information of different resolution in the wavelet domain, and preserve them with higher priority to realize high fidelity compression and resolution enhancement. Experiments are conducted on a real remote sensing image, and the results show that the algorithm can not only get high fidelity compression, at least 2dB improved, but also can enhance the resolution of reconstructed image. For the application of edge detection, the combined processing algorithm gets higher detection rate than Part 1 algorithm and Jasper algorithm. Furthermore, its detection rate has remained at least 90 percent in different compression ratio and the detected edge is more continuity.Finally, based on the theoretical algorithm study, according to the design idea of miniaturization, integration, hardware modules and software modules on multiple levels, a hardware system for RSI compression using ADV202,DSP and FPGA is designed and implemented. The working status shows that, the system can meet each design specifications and has good compressed effect, stable performance, high integration and wide applications. The system also provides a new idea for employing JPEG2000 standard in the satellite RSI compression.