Research On Hyperspectral Image Compression Based On JPEG2000

With higher ground resolution and spectrum resolution, hyperspectral remote sensing produce hyperspectral images, which contain not only spacial position and shape information but also spectral information. They are wildly used in resource exploration, environmental inspection, military detection, geographic prospecting, and so on. However, each pixel of hyperspectral image has tens or hundreds of magnitude values corresponding its wave bands, so the enormous data must be compressed. On the other hand, on the advantages of the current still image compression standard JPEG2000, it is widely used in remote sensing image compression.A compression scheme fit for interferential image sequence is proposed in the paper, which reduces correlation among image sequence by DCT, then compresses the eigenvalue images using the core algorithm of JPEG2000, and the region with interferential stripes are encoded as ROI.In this paper, the developing course of image compression technologies and various of current encoding methods are summarized, and hyperspectral remote sensing technology is introduced. In succession, every functional module of JPEG2000 encoder is explained in detail and the wavelet transform technology as well as the ROI encoding methods based on bit plane is introduced as emphases. On these bases, the whole compression encoding scheme is proposed, furthermore, the results of C++ emluator for this scheme are shown and analyzed. According to the experimental results, the scheme attains good quality of reconstructed images. Especially, the performance of reconstructed spectral images and curves are obviously improved.