Machine Learning-Based Fast Intra Mode Decision for HEVC Screen Content Coding via Decision Trees

Kuang, Wei; Chan, Yui-Lam; Tsang, Sik-Ho; Siu, Wan-Chi

doi:10.1109/tcsvt.2019.2903547

Cited by 42 publications

(37 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Besides, another set of hand-crafted rules are derived to skip partial PUs to further reduce the complexity of the re-encoding part. Although it shows better performance than the aforementioned fast SCC encoder techniques [24]- [28], it still leaves large room for further improvement. Since only limited hand-crafted rules are derived in [29], it may not handle the complicated situation in SCC transcoding well.…”

Section: Introductionmentioning

confidence: 98%

“…To reduce the computational complexity of SCC bitrate transcoding, one solution is to replace the original SCC encoder in CBFT by various fast SCC encoders [24]- [28]. In [24], a fast CU size decision method was proposed for stationary CUs in screen content videos.…”

Section: Introductionmentioning

confidence: 99%

“…Both PLT and IBC modes are searched for SCBs, while only Intra mode is checked for NIBs with N×N PUs. In [28], a sequential arrangement of decision trees was proposed to directly make mode classification for incoming CUs. The encoding time is further reduced since the different decision for PLT and SCB can be considered for a SCB.…”

Section: Introductionmentioning

confidence: 99%

“…Although the complexity of the SCC encoder is reduced by these techniques, the reduction is still limited. First, the fast prediction rules in [24]- [28] rely on the assumption that the computer-generated SCBs are noisefree, but this assumption does not hold for decoded videos in the CBFT due to the lossy encoding and decoding. Second, they only utilize the information from the encoder side for computational complexity reduction but do not consider the information from the decoder side.…”

Section: Introductionmentioning

confidence: 99%

“…To imitate the optimal mode selection in the original re-encoding part, CNN-BRT involves a loss function that considers both the sub-optimal modes and the final optimal modes. The difference between our contributions and the related schemes can be summarized as 1) unlike the fast SCC encoders in [24]- [28] that only consider information from the encoder side, CNN-BRT additionally utilizes information from the decoder side. 2) Compared with the fast SCC bitrate transcoder [29], CNN-BRT employs extensive learnable parameters to replace the limited hand-crafted rules.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Efficient Intra Bitrate Transcoding for Screen Content Coding Based on Convolutional Neural Network

Chan

Tsang

2019

IEEE Access

Self Cite

View full text Add to dashboard Cite

The Screen Content Coding (SCC) extension of High Efficiency Video Coding (HEVC) is developed to improve the coding efficiency of screen content videos. To meet the diverse network requirement of different clients, bitrate transcoding for SCC is desired. This problem can be solved by a conventional brute-force transcoder (CBFT) which concatenates an original decoder and an original encoder. However, it induces high computational complexity associated with the re-encoding part of CBFT. This paper presents a convolutional neural network based bitrate transcoder (CNN-BRT) for SCC. By utilizing information from both the decoder side and the encoder side, CNN-BRT makes a fast prediction for all coding units (CUs) of a coding tree unit (CTU) in a single test. At the decoder side, decoded optimal mode maps that reflect the optimal modes and CU partitions in a CTU is derived. At the encoder side, the raw samples in a CTU are collected. Then, they are fed to CNN-BRT to make a fast prediction. To imitate the optimal mode selection in the original re-encoding part, CNN-BRT involves a loss function that takes both of the sub-optimal modes and the final optimal modes into consideration. Compared with the HEVC-SCC reference software SCM-3.0, the proposed CNN-BRT reduces encoding time by 54.86% on average with a negligible Bjøntegaard delta bitrate increase of 1.01% under all-intra configuration. INDEX TERMS Transcoding, screen content coding (SCC), fast algorithm, convolutional neural network.

show abstract

Section: Introductionmentioning

confidence: 98%