Research On The Key Technologies Of Spaceborne Remote Sensing Image Real-time Compression Equipment Based On JPEG2000

Spaceborne remote sensing image compression technology, which benefits many military applications, is one of the key technologies in the aerospace sensing area. Among all types of image compression algorithms, JPEG2000 is the state-of-the-art image compression standard which has many advantages, such as high compression efficiency and high image fidelity. However, because of its high computational complexity, real-time hardware implementation becomes the biggest problem when using JPEG2000 in the aerospace sensing area. Supported by the military project"Development of××optical imaging satellite"and the 863 National High Technology Research and Development Program"××image data compression technology", this dissertation concentrates its attention on the key technologies of remote sensing image compression system based on JPEG2000. With the purpose of developing a remote sensing image compression equipment with own property right, the dissertation studies the high-quality and low-complexity fast algorithm for Discrete Wavelet Transform (DWT), the improved EBC algorithm suited for DSP parallel implementation, the new approach for remote sensing image quality evaluation, and the development of remote sensing image compression equipment.The main contents and contributions of this dissertation are as follows:1. Aiming at the shortcut of computation complexity of DWT for implementation in the real-time remote sensing image compression, the fixed point algorithms and fast implementation methods are studied. A fixed point algorithm based on LS9/7 is proposed firstly. Compared with the fixed point algorithm based on CDF9/7, the proposed algorithm greatly lowers the computational complexity, while still achieves the same image fidelity as the float point algorithm based on CDF9/7. Then, based on post scaling transform, a modified Spatial Combinative Lifting Algorithm (SCLA) is designed to further lower the computational complexity. At last, for the purpose to implement JPEG2000 with high speed, a quantification-removed method is proposed to simplify the JPEG2000 image compression algorithm.2. Real-time implementation of EBC algorithm is a difficult problem when using JPEG2000 in remote sensing image compression equipment. Aiming at the shortcut of the EBC framework with many loops which is not suitable for the establishment of DSP software pipeline, an optimization framework for EBC algorithm is proposed. The original bit-plane coding module is partitioned into bit-plane scanning module and arithmetic coding module. And the multiple loops of the bit-plane scanning module are unwrapped with the context combining. In the optimization framework, the arithmetic coding module and the storage structure of the wavelet coefficients are modified. By the optimization, the software pipeline of the arithmetic coding can be established which leads to a great acceleration of the EBC implementation speed.3. For the software pipeline of the bit-plane scanning module can not established even in the optimization framework, an improved bit-plane scanning method based on lone sign coding is proposed. The sign coding is separated from the bit-plane scanning module and implemented ahead. The improved method simplifies the framework of the bit-plane scanning module and speed up the implementation greatly.4. To further improve the coding efficiency of EBC algorithm, a no-time spending data moving strategy based on parallel mechanism is proposed. It is implemented by splitting the EBC algorithm and adapting EDMA transport method, which make the EBC algorithm and data moving between the FPGA and DSP parallel executing. With this new data moving strategy, the time speeding on data moving can be ignored which accelerate the EBC implementation speed.5. Taking into account the trait of the remote sensing image and the human vision characteristics, an objective evaluation model with a new evaluation index, named edge weighted peak signal noise ratio, is proposed. The proposed objective evaluation model pays more attention to the edge distortion of the remote sensing image. Extracting the edges from the image, a weighting factor corresponding with the edge zone union of the image is introduced into the model to ensure the edge distortion influence the result of the evaluation more. The test results show that the proposed objective image quality evaluation method is more consistent with the subjective method than PSNR based one.6. Inspired by the pipeline and Single Program Multiple Data (SPMD) concepts, a hardware platform based on FPGA and parallel multiple DSPs is designed. A JPEG2000 based remote sensing image compression equipment prototype with own property right is developed based on this platform. Then the prototype is tested. It is shown that the system can gain the requiring image quality under specific real-time need and power requirement. Now it has been entered the phase of sample development period which lays the foundation for the successful development of the optical imaging satellite.