Remote-Sensing Image Compression Using Priori-information and Feature Registration

In this paper, we focus on a high performance compression scheme for remote-sensing images, which is essential due to limited transmission bandwidth while explosively growing remote-sensing image data size. First, on the basis of intra-image spatial redundancy removal, which is used by JPEG 2000 and CCSDS, priori-information is introduced to eliminate temporal redundancy between historical and newly-captured images, at the same time. Second, feature registration technique is applied rather than motion estimation and compensation which is used in HEVC, to deal with the long-range non-linear correlation of remote-sensing image series. Numerical simulation results show that the proposed scheme outperforms JPEG 2000 and JPEG by over 1.37 times for lossless compression, and presents a 5 dB PSNR gain over JPEG 2000 and HEVC for lossy compression.