JISE


  [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]


Journal of Information Science and Engineering, Vol. 39 No. 2, pp. 339-352


Fast Intra Coding Partition Decisions for Feature Video Coding


JIUNN-TSAIR FANG1,+, YU-LIANG TU2 AND PAO-CHI CHANG2
1Department of Electronic Engineering
Ming Chuan University
Taoyuan City, 333 Taiwan
E-mail: fang@mail.mcu.edu.tw

2Department of Communication Engineering
National Central University
Taoyuan City, 320 Taiwan


Future video coding (FVC) can support 4K high-resolution videos to replace the previous coding standard, high efficiency video coding (HEVC). In particular, FVC adopts quadtree with binary tree (QTBT) to improve coding efficiency; however, encoding time increases heavily. Thus, we propose fast partition algorithms for FVC intra coding. Fast partition algorithms include spatial correlation method and deep learning method. The spa-tial correlation method uses gradient variances of pixels to determine early skip and early termination for QT depths 0 and 1, respectively. The deep learning method uses convolu-tional neural networks (CNNs) to predict the QTBT coding structure at QT depth 2 with its corresponding coding trees. Experimental results show that the proposed method can reduce encoding time by 12.49% on average but increases Bjontegaard delta bit rate (BDBR) by about 0.54%.


Keywords: future video coding, intra coding, quadtree with binary tree, coding unit, convolutional neural networks

  Retrieve PDF document (JISE_202302_08.pdf)