Advanced screen content coding using color table and index map.

Area of Science:

• Digital image processing
• Video compression algorithms
• Computer vision

Background:

• High-Efficiency Video Coding (HEVC) extensions (HEVC-RExt) offer advanced video compression.
• Screen content, characterized by text and graphics, presents unique compression challenges due to high redundancy.
• Conventional prediction methods in HEVC-RExt may not fully exploit screen content-specific redundancies.

Purpose of the Study:

• To develop and evaluate an advanced screen content coding solution.
• To enhance the coding efficiency of HEVC-RExt for screen content.
• To exploit redundancy in screen content using a novel color table and index map (cTIM) method.

Main Methods:

• Implementation of the color table and index map (cTIM) method within HEVC-RExt as an intra-coding tool.
• Formation of a color table from major colors and translation of pixel blocks to an index map for each coding unit.
• Utilization of 1D or hybrid 1D/2D string matching for index map compression.
• Development of color table merge, inter-table color sharing, and intra-table color differential predictive coding to reduce signaling overhead.

Main Results:

• Demonstrated significant coding efficiency improvements over conventional HEVC-RExt.
• Achieved bit rate reductions of 26%, 18%, and 15% in lossless mode for all intra, random access, and low-delay settings, respectively.
• Obtained Bjontegaard Delta-rate improvements of 23%, 19%, and 13% in lossy mode for typical screen content.

Conclusions:

• The proposed cTIM method effectively exploits screen content redundancy.
• The cTIM scheme offers substantial coding efficiency gains for both lossless and lossy compression of screen content.
• The algorithm is efficient and provides significant improvements compared to existing HEVC-RExt methods.