In my current postdoctoral and previous PhD research undertakings, there is a focused concentration on the development of perceptually optimised video compression algorithms (i.e., visually lossless coding) for the HEVC standard. HEVC (ISO/IEC 23008–2, ITU-T H.265) is a video compression platform that has been internationally standardised by JCT-VC (VCEG of ITU-T and also MPEG of ISO and IEC). With application to various types of YCbCr and RGB Big Data, I am presently developing visually lossless compression mathematical models and algorithms, for which I utilise both MATLAB and C++. The raw YCbCr and RGB video data to which the proposed algorithms can be applied includes medical image data and various types of screen content including camera-captured content and animated content. My postdoctoral work follows on from the research that I conducted during the PhD degree in Computer Science.
Visually Lossless Coding for the HEVC Standard: Efficient Perceptual Quantisation Contributions for HEVC (PhD Completed: September, 2017)
In relation to my PhD research and thesis, four perceptually adaptive quantisation techniques are proposed for the HEVC standard. These contributions are designed to maximise the levels of compression which can be applied to uncompressed video data (i.e., raw YCbCr sequences) — of various bit depths, resolutions and sampling ratios — without incurring a perceptually discernible loss of reconstruction quality in the coded video sequence. This potentially gives rise to considerable improvements in terms of bitrate reductions, as measured in kilobits per second; furthermore, by virtue of the nature of the proposed methods, computational complexity is typically not increased. The following coding efficiency and visual quality metrics: Bjøntegaard Delta Rate (BD-Rate), PSNR and SSIM, in addition to subjective assessments that follow the principles of ITU-T P.910 "Subjective Video Quality Assessment Methods", are utilised to evaluate each contribution. The proposed perceptual quantisation methods are as follows:
1) High Bit Depth Capable and 4:4:4 Capable JND-Based Coding Block (CB)-Level Perceptual Quantisation;
2) Coding Block (CB)-Level Full Colour Perceptual Quantisation for 4:4:4 Video Data;
3) Coding Unit (CU)-Level Cross-Colour Channel Perceptually Adaptive Quantisation;
4) Transform Coefficient-Level Perceptual Quantisation.
Lee Prangnell and Victor Sanchez, "JND-Based Perceptual Video Coding for 4:4:4 Screen Content Data in HEVC", IEEE International Conference on Acoustics, Speech and Signal Processing, Calgary, Alberta, Canada, 2018, DOI: 10.1109/ICASSP.2018.8462327. (Preprint PDF).
Lee Prangnell, "Visually Lossless Coding in HEVC: A High Bit Depth and 4:4:4 Capable JND-Based Perceptual Quantisation Technique for HEVC", Elsevier Signal Processing: Image Communication Journal, April 2018, DOI: 10.1016/j.image.2018.02.007. (Preprint PDF).
Lee Prangnell, Miguel Hernández-Cabronero and Victor Sanchez, "Coding Block-Level Perceptual Video Coding for 4:4:4 Data in HEVC", IEEE International Conference on Image Processing, Beijing, China, 2017, DOI: 10.1109/ICIP.2017.8296730. (Preprint PDF).
Lee Prangnell, Miguel Hernández-Cabronero and Victor Sanchez, "Cross-Color Channel Perceptually Adaptive Quantization for HEVC", IEEE Data Compression Conference, Snowbird, Utah, USA, 2017, DOI: 10.1109/DCC.2017.66. (Preprint PDF).
Lee Prangnell and Victor Sanchez, "Adaptive Quantization Matrices for HD and UHD Resolutions in Scalable HEVC", IEEE Data Compression Conference, Snowbird, Utah, USA, 2016, DOI: 10.1109/DCC.2016.47. (Preprint PDF).
Lee Prangnell, Victor Sanchez and Rahul Vanam, "Adaptive Quantization by Soft Thresholding in HEVC", IEEE Picture Coding Symposium, Cairns, Queensland, Australia, 2015, DOI: 10.1109/PCS.2015.7170042. (Preprint PDF).
Lee Prangnell, "Visually Lossless Coding for the HEVC Standard: Efficient Perceptual Quantisation Contributions for HEVC", PhD Thesis, Department of Computer Science, University of Warwick, September 2017 (PDF).