JPEG2000 And Its Application In Compression Of Hyperspectral Remote-sensing Images

With the development of remote sensing technology and the appearance of the remote sensing device for plenty of remote sensing data, compression of hyperspectral images is of more and more importance. As the original data is needed sometimes, the lossless compression of hyperspectral images is necessary. JPEG2000 is the newest still image compression standard. It gives up the block coding style based DCT employed by JPEG. It employs multi-analyse based on wavelet transform and improves performance of image compression. In this paper we expand on the basic algorithm and key technology of JPEG2000, and compress hyperspectral remote sensing images using JPEG2000 algorithm.At first the significance of image compression is discussed, the existing algorithm of image compression and the compression technology of hyperspectral image is reviewed. Secondly, the theories basis of JPEG2000' s basic algorithm is introduced from the viewpoint of information theory. The key algorithm including lifting scheme of wavelet and EBCOT is described. Then, the compression algorithm of JPEG2000 standard is analyzed detailed, including DC level shifting, component transformation, wavelet transformation, code block bit-plane coding, layers and packets, especially the wavelet transformation, three fraction bit-plane partition and layers. At last, realizes a hybrid algorithm of using JPEG2000 and predictive coding. Inter-band redundancy is removed by adaptive prediction, and the algorithm of block partition prediction is proposed because of the complexity of the adaptive prediction. The residual images is compressed by EBCOT and create the JPEG2000 code stream. The experiment results show the compression radio is from 2.54 to 3. 3 for the data of 64 band hyperspectral images and near 2 for the 6 band TM multispectral images.