Research And FPGA Implementation Of Adaptive Lossless Compression For Multispectral Image

With the continuous development of remote sensing technology and multi-spectral imaging technology, the spatial resolution, time resolution and spectral resolution of the multispectral remote sensing image are increasing. As the information is expanding, higher requirements for the storage, transmission and use are put forward. Moreover, the multispectral image is mainly obtained by remote sensing satellite, it costs high and needs to save for long time. As a result, it generally require lossless compression. The main research of this paper is to design a lossless compression algorithm of multispectral remote sensing image by FPGA.The characteristics of multispectral remote sensing image data are analyzed in this paper, different compression methods are respectively designed for the redundancy in the spectral and among the spectral. For the redundancy in the spectral, the kernel LOCO-1(Low Complexity Lossless Compression) of the JPEG-LS (Lossless and near-lossless compression Standard of JPEG) is chosen as predictor to realize the decorrelation in spectral. For the redundancy among the spectral, the RCT(Reversible Component Transform) transform for hardware implementation to realize the decorrelation among spectral, as the RCT is reversible and it is Suitable for lossless compression. Combining two related modules, a quick and efficient approach is adopted to eliminate the redundancy in the spectral and among the spectral of the multispectral remote sensing image.The block coding can effectively improve the coding efficiency and reduce the error diffusion. There is no need to introduce external RAM in the process of encoding.In the entropy coding, we improve the algorithm of RICE entropy encoder based on the characteristic of structure of the FPGA (Field Programmable Gate Array). Through simple the accumulator, the selector and Golomb encoder, efficient compression could be implemented. The entropy coding is more simple, hardware resources are more saved. Meanwhile, in order to improve the coding efficiency and the utilization of hardware resources, multichannel parallel and pipeline coding mode are developed.