Quality improvement of tone mapped images by TMQI-II based optimization for the JPEG XT standard

Abstract

This paper proposes a quality improvement method of tone mapped images including compression distortion for the JPEG XT standard, which is a new compression standard for high dynamic range (HDR) images. HDR images are generally required to be mapped to low dynamic range (LDR) ones due to the limitation of display devices. Furthermore, the HDR ones include some compression distortion to be efficiently stored, in most cases. However, conventional tone mapping operations have not considered the effect of the distortion. We apply an iterative gradient ascend algorithm for improving the structural fidelity, which is based on the improved tone mapped image quality index (TMQI-II), under the use of the JPEG XT standard. Decompressed algorithm and the results are compared with those of original HDR ones. The experiment confirms that the images with better quality than images without optimization are provided by the proposed method and the quality is close to that of ones generated from the original HDR image without compression distortion.