Based On The Adv212 Spectral Data Compression And Decompression

Hyperspectral remote sensing combines remote sensing images andspectra èunifies the logical thinking and thinking in images of humanepistemology and greatly enhance the human cognitive abilities about theobjective world. Imaging spectrometer can get hundreds of bands ofcontinuous images, and we can extract a spectral curve from each imagepixel. Unusually large amount of spectral data has led to greatdifficulties for the data storage and transmission. It directly affectsthe application of the imaging spectrometer. So the research on spectraldata compression technology is particularly important.JPEG2000is an upgraded version of JPEG and is the latestinternational standard for static image compression. The compressionratio is higher than JPEG about30%and the compression performance isgood at low bit rate. it provides both lossless and lossy compression,supports coding the region of interest and can randomly access and processthe code stream. It has good error robustness, gradually transmits SNR(Signal to Noise Ratio) and resolution and has open framework structure.There are many ways to achieve JPEG2000coding. This paper use FPGAequipped with dedicated image compression chip ADV212, whose compressionstandard is JPEG2000, to come to the system of spectral datacompression.In this way, we have not to spend a lot of time optimizingthe algorithm. The system implements simply and its technology is mature,reliable and stable because of using a dedicated image compression chip.Our design, targeting on the mass Spectral data, and after we exploredthe spectral data compression, comes to the final scheme of imagecompression and decompression based on ADV212, whose compression standardis JPEG2000, and MicroBlaze soft-core processor from Xilinx. The MicroBlaze soft-core processor completed image compression anddecompression through configuration of the ADV212via the I2C bus. Thehardware compression system was then verified on DMD-based HadamardTransform Imaging Spectrometer.The results show that our design reaches the requirements of real-timespectral data compression, and is simpler and take less time to finish.In addition, compared with software, it improves the overall systemperformance.