A hyperspectral image compression algorithm based on wavelet transformation and fractal composition (AWFC)  

作　　者：HU Xingtang ZHANG Bing ZHANG Xia ZHENG Lanfen TONG Qingxi 

机构地区：[1]State Key Laboratory of Remote Sensing Science, Jointly Sponsored by the Institute of Remote Sensing Applications of Chinese Academy of Sciences and Beijing Normal University, Beijing, 100101, China State Key Laboratory of Remote Sensing Science, Jointly Sponsored by the Institute of Remote Sensing Applications of Chinese Academy of Sciences and Beijing Normal University, Beijing, 100101, China State Key Laboratory of Remote Sensing Science, Jointly Sponsored by the Institute of Remote Sensing Applications of Chinese Academy of Sciences and Beijing Normal University, Beijing, 100101, China State Key Laboratory of Remote Sensing Science, Jointly Sponsored by the Institute of Remote Sensing Applications of Chinese Academy of Sciences and Beijing Normal University, Beijing, 100101, China State Key Laboratory of Remote Sensing Science, Jointly Sponsored by the Institute of Remote Sensing Applications of Chinese Academy of Sciences and Beijing Normal University, Beijing, 100101, China

出　　处：《Science China(Technological Sciences)》2006年第z2期48-56,共9页中国科学（技术科学英文版）

摘　　要：Starting with a fractal-based image-compression algorithm based on wavelet transformation for hyperspectral images, the authors were able to obtain more spectral bands with the help of of hyperspectral remote sensing. Because large amounts of data and limited bandwidth complicate the storage and transmission of data measured by TB-level bits, it is important to compress image data acquired by hyperspectral sensors such as MODIS, PHI, and OMIS; otherwise, conventional lossless compression algorithms cannot reach adequate compression ratios. Other loss-compression methods can reach high compression ratios but lack good image fidelity, especially for hyperspectral image data. Among the third generation of image compression algorithms, fractal image compression based on wavelet transformation is superior to traditional compression methods,because it has high compression ratios and good image fidelity, and requires less computing time. To keep the spectral dimension invariable, the authors compared the results of two compression algorithms based on the storage-file structures of BSQ and of BIP, and improved the HV and Quadtree partitioning and domain-range matching algorithms in order to accelerate their encode/decode efficiency. The authors' Hyperspectral Image Process and Analysis System (HIPAS) software used a VC++6.0 integrated development environment (IDE), with which good experimental results were obtained. Possible modifications of the algorithm and limitations of the method are also discussed.Starting with a fractal-based image-compression algorithm based on wavelet transformation for hyperspectral images, the authors were able to obtain more spectral bands with the help of of hyperspectral remote sensing. Because large amounts of data and limited bandwidth complicate the storage and transmission of data measured by TB-level bits, it is important to compress image data acquired by hyperspectral sensors such as MODIS, PHI, and OMIS; otherwise, conventional lossless compression algorithms cannot reach adequate compression ratios. Other loss-compression methods can reach high compression ratios but lack good image fidelity, especially for hyperspectral image data. Among the third generation of image compression algorithms, fractal image compression based on wavelet transformation is superior to traditional compression methods, because it has high compression ratios and good image fidelity, and requires less computing time. To keep the spectral dimension invariable, the authors compared the results of two compression algorithms based on the storage-file structures of BSQ and of BIP, and improved the HV and Quadtree partitioning and domain-range matching algorithms in order to accelerate their encode/decode efficiency. The authors' Hyperspectral Image Process and Analysis System (HIPAS) software used a VC++6.0 integrated development environment (IDE), with which good experimental results were obtained. Possible modifications of the algorithm and limitations of the method are also discussed.

关 键 词：WAVELET transformation  FRACTAL coding  IMAGE compression  HYPERSPECTRAL image  HIPAS. 

分 类 号：TH[机械工程]