Research And Implementation Of Ccsds Lossless Compression Algorithm For Hyperspectral Image

With the advancement of aerospace science and technology,the resolution of spectral images has been increasing.The amount of information in optical remote sensing images has increased exponentially.The limited transmission bandwidth and hardware resources of satellite-borne devices cannot meet the demand for massive amounts of data.Therefore,how to effectively compress data is particularly important.Based on CCSDS123.0-B-1 hyperspectral image lossless compression algorithm standard.Under the project requirements for lossless compression of a certain model of hyperspectral image data,the algorithm is improved and the hardware implementation is optimized.The algorithm is implemented on the Xilinx K7 series chips.This thesis first analyzes the hyperspectral image compression algorithm.According to the algorithm principle,if a certain internal pixel is compressed incorrectly,all subsequent compressed data will be wrong.In order to avoid error accumulation,it chooses 32 rows of image data as a data cube and after each data cube is compressed,the compressor is initialized,which eliminates the possibility of infinite error transmission and effectively improves the error resistance rate of the compressor.Secondly,this thesis presents the FPGA implementation of high-speed image compression system.The compression system consists of predictor and encoder.In the design of the predictor,since the weight coefficient is needed to calculate the predicted value,the predicted value needs to be used to update the weight coefficient after the predicted value is obtained.When this algorithm structure is mapped to the hardware implementation,a feedback structure will inevitably be generated.It will affect the implementation speed of the algorithm on the hardware.In the design,this thesis adopts dual clock and insert pipeline solution to solve the feedback problem in the system,shorten the calculation delay inside the feedback loop,and improve the calculation speed of the algorithm.The system input image data transmission bandwidth can reach 960Mbps,which can meet the actual demand of the project.In the design of the encoder,when encoding the predicted value,it was found that the Golomb Rice encoding method used would produce a large number of consecutive ’0’ codes.The existence of this situation affects the image compression rate.Therefore,when the system is designed,the mapping value is compared to the threshold before encoding.If a large number of ’0’s are found after encoding,the encoding is not performed and the original data is directly written.Finally,this thesis builds a system simulation verification platform,and uses the host computer to simulate and verify the image compression system.The test data used in this thesis are all from the actual image data produced by the hyperspectral camera.The continuous data of different widths and spectral bands are tested.While comparing the result of hardware compression with the result of host computer software compression,the compressed data was decompressed and compared with the original data to verify the correctness of the compression system design.The verification results show that the hyperspectral image compression system designed in this paper meets the requirements of aerospace missions in terms of compression performance,safety and reliability.