2020/06/11 by Luka Murn, Saverio Blasi, Alan F. Smeaton +2
Computer Science · Engineering · #Advanced Image Processing Techniques #Advanced Vision and Imaging #Artificial neural network #Coding (social sciences) #Coding tree unit #Convolutional neural network #Data compression #Interpolation (computer graphics) #Motion compensation #Motion interpolation #Video Coding and Compression Technologies #cs.CC #cs.CV #cs.LG #cs.MM #eess.IV
paper · pdf · doi:10.1109/icip40778.2020.9191193
published as 2020 IEEE International Conference on Image Processing (ICIP), 2020, pp. 798-802 · 27th IEEE International Conference on Image Processing, 25-28 Oct 2020, Abu Dhabi, United Arab Emirates
arxiv created 2020/06/11 · openalex created_date 2020/06/19 · openalex publication_date 2020/09/30 · arxiv updated 2021/05/28 · openalex updated_date 2026/08/05
Deep learning has shown great potential in image and video compression tasks. However, it brings bit savings at the cost of significant increases in coding complexity, which limits its potential for implementation within practical applications. In this paper, a novel neural network-based tool is presented which improves the interpolation of reference samples needed for fractional precision motion compensation. Contrary to previous efforts, the proposed approach focuses on complexity reduction achieved by interpreting the interpolation filters learned by the networks. When the approach is implemented in the Versatile Video Coding (VVC) test model, up to 4.5% BD-rate saving for individual sequences is achieved compared with the baseline VVC, while the complexity of learned interpolation is significantly reduced compared to the application of full neural network.