2021/02/08 by Angeliki V. Katsenou, Angeliki Katsenou, Katsenou, Angeliki V. +5 · 2 citations
Computer Science · Engineering · Mathematics · #Advanced Data Compression Techniques #Artificial intelligence #Coding (social sciences) #Computer science #Computer vision #Data compression #Distortion (music) #Encoding (memory) #FOS: Electrical engineering #Image (mathematics) #Image and Video Processing (eess.IV) #Image and Video Quality Assessment #Mathematics #Multiview Video Coding #Pattern recognition (psychology) #Quantization (signal processing) #Rate–distortion optimization #Rate–distortion theory #Statistics #Texture (cosmology) #Uncompressed video #Video Coding and Compression Technologies #Video processing #Video quality #Video tracking #eess.IV #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2102.04167
17 pages
arxiv created 2021/02/08 · openalex publication_date 2021/02/08 · arxiv updated 2021/02/09 · openalex created_date 2022/07/25 · openalex updated_date 2026/08/06
Encoding textural content remains a challenge for current standardised video\ncodecs. It is therefore beneficial to understand video textures in terms of\nboth their spatio-temporal characteristics and their encoding statistics in\norder to optimize encoding performance. In this paper, we analyse the\nspatio-temporal features and statistics of video textures, explore the\nrate-quality performance of different texture types and investigate models to\nmathematically describe them. For all considered theoretical models, we employ\nmachine-learning regression to predict the rate-quality curves based solely on\nselected spatio-temporal features extracted from uncompressed content. All\nexperiments were performed on homogeneous video textures to ensure validity of\nthe observations. The results of the regression indicate that using an\nexponential model we can more accurately predict the expected rate-quality\ncurve (with a mean Bj ontegaard Delta rate of 0.46% over the considered\ndataset) while maintaining a low relative complexity. This is expected to be\nadopted by in the loop processes for faster encoding decisions such as\nrate-distortion optimisation, adaptive quantization, partitioning, etc.\n