Hyperspectral Images Compression Based on Independent Component Analysis ROI-based compression algorithm for hyperspectral images

This paper addresses the problem of lossy compression for hyperspectral images and presents an efficient compression algorithm based on FastICA. Firstly, an efficient algorithm for segmentation of hyperspectral images is proposed. Secondly, based on the targets, a lossy compression based on ROI (Region of Interest) is proposed for hyperspectral compression, which employs KLT(Karhunen-Loeve transform) to remove the spectral correlation and DWT(Discrete Wavelet Transform) to remove the spatial correlation. Moreover, scaled-based shift algorithm is used to shift the wavelet coefficients of the interested targets; Finally, SPIHT(Set Partitioned In Hierarchical Tree) algorithm is used to compress each band. Experimental results show that the proposed algorithm can efficiently protect the target information of hyperspectral images even if at low bitrates.