Wavelet-based Compression for Multispectral Imagery
基于小波变换的多光谱图像压缩方法

For the multispectral image (MSI) data, there are two types of redundancy spatial redundancy and spectral redundancy. In this paper, we classified the spectral redundancy into two categories spectral statistical redundancy and spectral structural redundancy. The former is based on the spectral resolution. The higher spectral resolution the more redundancy. The latter is caused by the same imaging objects of all bands images, it is essentially based on the arrangement of earth objects. Essentially, the two types of redundancy are different. Here, we proposed a lossy compression technique based on wavelet transformation Share Significance Map Wavelet Transform (SSMWT). With this technique, zerotree coding was used in compression of MSI, and we only need to create one significance map for all bands of images in MSI, is the light of structure correlation between all bands of images after WT, and then to remove spatial correlation and spectral structure correlation. Combined with K-L transformation, the spectral statistic correlation of MSI can be removed. The experiments have shown the efficiency of this technique.