作者: Mario Vranješ , Ratko Grbić , Matteo Brisinello , Tomislav Jurić
DOI: 10.1007/S11554-020-01065-9
关键词: Data compression 、 Lossy compression 、 Quality of experience 、 Blocking (statistics) 、 Artifact (error) 、 Computer vision 、 Artificial intelligence 、 Video quality 、 Computer science 、 Frame rate 、 Visibility (geometry)
摘要: During lossy video compression, a blocking artifact is introduced into video. It can significantly reduce the quality of compressed video, and consequently end-user Quality Experience. Hence, it very important to measure monitor videos delivered end-user, which be performed only by using no-reference (NR) approach, without original signal available. Many NR assessment (VQA) metrics are based on measurements different artifacts visibility, then used as an input for objective VQA metric. Thus, reliable detection their visibility estimation first step in process. In this paper, new algorithm Real-Time Blocking Visibility Estimation (RT-BVE) frames proposed. The performance proposed RT-BVE compared two freely available algorithms, i.e., BBT-BDA MSU Blocking, from well-known databases (156 sequences total more than 48,000 frames). While achieves comparable algorithms MPEG-2 videos, outperforms both H.264 videos. results show that precisely estimate according compression standards fixed set parameter values. Additionally, suitable usage real-time applications since capable processing approximately 100 Full HD per second current midrange × 86–64 platform.