Alogorithms for fast implementation of high efficiency video coding

• मुख्य लेखक: Jaja, Edward Tamunoiyowuna
• स्वरूप: थीसिस
• भाषा: अंग्रेज़ी
• प्रकाशित: 2017
• विषय: TK Electrical engineering. Electronics Nuclear engineering
• ऑनलाइन पहुंच: http://eprints.utm.my/84023/2/EdwardTamunoiyowunaJajaPFKE2017.pdf

Recently, there is higher demand for video content in multimedia communication, which leads to increased requirements for storage and bandwidth posed to internet service providers. Due to this, it became necessary for the telecommunication standardization sector of the International Telecommunication Union (ITU-T) to launch a new video compression standard that would address the twin challenges of lowering both digital file sizes in storage media and transmission bandwidths in networks. The High Efficiency Video Compression (HEVC) also known as H.265 standard was launched in November 2013 to address these challenges. This new standard was able to cut down, by 50%, on existing media file sizes and bandwidths but its computational complexity leads to about 400% delay in HEVC video encoding. This study proposes a solution to the above problem based on three key areas of the HEVC. Firstly, two fast motion estimation algorithms are proposed based on triangle and pentagon structures to implement motion estimation and compensation in a shorter time. Secondly, an enhanced and optimized inter-prediction mode selection is proposed. Thirdly, an enhanced intra-prediction mode scheme with reduced latency is suggested. Based on the test model of the HEVC reference software, each individual algorithm manages to reduce the encoding time across all video classes by an average of 20-30%, with a best reduction of 70%, at a negligible loss in coding efficiency and video quality degradation. In practice, these algorithms would be able to enhance the performance of the HEVC compression standard, and enable higher resolution and higher frame rate video encoding as compared to the stateof- the-art technique.