Nearly lossless HDR images compression by background image segmentation

Abstract

High-Dynamic Rang and high spatial resolutions image, such as 16bits X-ray image, is required in security and medical applications. However, transporting this kind of the image costs a lot in network bandwidth, storage capacity and transmission time. To solve the problem, efficient compression technology is raised. In this paper, we propose a nearly lossless compression algorithm with background segmentation for 16 bits X-ray image. Background and foreground of image are separated with a threshold calculated through detecting background peak of image gray histogram. Background pixels connected to image border are segmented as background by two pass run-length labeling. Then image background is compressed with Run-Length-encoding (RLE) and the foreground is encoded via Lempel-Ziv-77 (LZ77). To validate the proposed algorithm, we compare it with state-of-art compression algorithms LZ77, JPGE_LS, and arithmetic coding. Experimental results show that our algorithm obtains good performance both on compress ratio and speed with neglectable information loss.