Color Information Restoration by the Hue-Space Sparse Representations and Soft-Decision

点击翻译按钮获取中文摘要

This project aimed at the improvement in the restoration of color signals for digital imaging. A sparse representation was employed in the hue space. The reconstructed/decoded quality stays around 30dB in PSNR for still-imaging and video decoding of intra-frames, while the one-fifth data is enough to reproduce the quality level comparable to the H.264/AVC. In addition, a color-tone similarity index was invented and was introduced into a new partial-reference (PR) image quality assessment (IQA). The PR IQA is competitive to the full-reference IQA in spite of much less data.