Lossless Bit Depth Scalable Coding for Floating Point Images

Chew Yin Ping, Takuya Shibata, Masahiro Iwahashi, Hitoshi Kiya · 2013

This paper presents a two-layer lossless bit depth scalable coding for RGB color components of high dynamic range (HDR) images. We introduce a two-stage tone mapping, which is composed of a reversible logarithmic mapping and its compensation, to produce a tone mapped LDR image in the base layer. Adding bit stream in the enhancement layer, which carries difference between the original HDR image and its approximation, the original HDR image is reproduced without any loss. The reversible logarithmic mapping is utilized to reduce bit depth of the enhancement layer for lossless coding. We have confirmed that the proposed method significantly reduces bit depth and bit rate of HDR color images in the enhancement layer.

Read the paper · More papers on PaperTik