On visual masking estimation for adaptive quantization using steerable filters
作者：
Highlights：
• Steerable features can effectively predict perceptual contrast masking effects.
• A reliable contrast masking threshold predictor is derived using steerable features.
• The predictions are applied for the adaptive quantization process of video encoding.
• The approach reduces compression artifacts at low bitrates.
• Achieves better rate–distortion tradeoff than variance based adaptive quantization.
摘要
•Steerable features can effectively predict perceptual contrast masking effects.•A reliable contrast masking threshold predictor is derived using steerable features.•The predictions are applied for the adaptive quantization process of video encoding.•The approach reduces compression artifacts at low bitrates.•Achieves better rate–distortion tradeoff than variance based adaptive quantization.
论文关键词：Adaptive quantization,Steerable filtering,Visual masking,AV1
论文评审过程：Received 30 September 2020, Revised 1 March 2021, Accepted 9 April 2021, Available online 15 April 2021, Version of Record 23 April 2021.
论文官网地址：https://doi.org/10.1016/j.image.2021.116290
