Ultra-Low-Latency Video Coding with Reduced Frame Memory Structure for 4K/8K High-Resolution Video

• Published in: 2023 IEEE 12th Global Conference on Consumer Electronics (GCCE) pp. 852 - 853
• Main Authors: Yamaguchi, Mai, Yamakawa, Mai, Mochizuki, Seiji, Imamura, Kousuke, Kanamoto, Toshiki, Matsumura, Tetsuya
• Format: Conference Proceeding
• Language: English
• Published: IEEE 10-10-2023
• Subjects: autonomous driving; high resolution video; Image coding; Image quality; Image resolution; low latency; Scalability; Video coding; Virtual reality; virtual reality (VR); Visualization
• ISSN: 2693-0854
• DOI: 10.1109/GCCE59613.2023.10315450

In recent years, with the increasing demand for low-latency transmission of high-resolution video in various applications such as autonomous driving and virtual reality (VR), efficient low-latency encoding methods are needed. We previously reported a line-based video encoder for Full-HD with ultra-low-latency on the order of microseconds. This paper introduces a newly developed ultra-low-latency video encoding method designed to support 4K/8K high-resolution video. Our approach incorporates inter-frame prediction from the same position of the previous frame with the Reduced Frame Memory Structure (RFMS). This structure provides scalability while enhancing the compression ratio, even in situations with insufficient frame memory for larger 4K/8K video data. In addition, we apply a context-based adaptive VLC to further improve the compression ratio for 4K and 8K resolutions. These features achieve compression ratios of 10.19% and 9.20% at 40.82 dB and 40.63 dB of image quality for 4K and 8K, respectively.