The Optimization Of Parallel JPEG2000 Decompression Algorithm Based On Multi-core Processor

With the rapid increasing requirements in the development of Earth Observation Satellite, a large amount of Remote Sensing Image data needs to be processed, which poses a severe challenge to the development of the Ground Image Detection and processing system. Due to the features of high bitrate of data transmission channel, complexities of data processing, high calculation speed and so on, high speed decoding of large amount of Remote Sensing Image data increasingly attracts people’s attention. However, the standard compression algorithm is very complicated and time-consuming in decoding. In order to achieve highly efficient decoding, a real-time decoding processing system with high performance of panchromatic images in JPEG2000 standard which employs multi-core processors is developed in this thesis.Multi-core processor refers to a processor integrated with multiple complete computing engines, so that each engine can calculate parallel and independently, which largely increase the performance compared with the traditional single CPU platform. The multi-core processor used in this thesis is the 4 TILE-GX36 CPU platform of Tilera,and each TILE-GX36 CPU has 36 independent cores. In order to achieve the fast and reliable method of image decompression, JPEG2000 decoding algorithm is embedded into the TILE-GX36 platform. Then, we optimize the algorithms and hardware platform to efficiently solve the urgent problems of the low rate of real-time decompression in the current tasks.In conclusion, firstly, the history and the current situation of multi-core processors is discussed in the main text of this thesis, then a simple introduction is given to the core technology of multi-core processors. Besides, JPEG2000 decoding algorithm is discussed in detail. Finally, a high speed parallel implementation scheme of JPEG2000 decoding of Tilera platform based on multi-core processors is presented, and the decoding speed is increased by the optimization of algorithms and platforms. At last, performance tests are carried on based on the platform and the test results reach the target of correct and high-speed decoding.