Font Size: a A A

Research On Learning Based Efficient Transcoding

Posted on:2017-05-20Degree:MasterType:Thesis
Country:ChinaCandidate:Y Q X HuangFull Text:PDF
GTID:2428330590468250Subject:Electronic and communication engineering
Abstract/Summary:
High Efficiency Video Coding(HEVC),was developed as the next generation of video coding standard to replace H.264/AVC.Compared with H.264/AVC,HEVC is more flexible and high efficient,which aims to double the compression ratio of H.264/AVC.Based on the more flexible coding unit,more comprehensive intra and Inter prediction and some new techniques,HEVC improves a lot in performance.Due to the increasing computation complexity,reducing the the compelixty of HEVC encoder based on ensuring the efficiency has become an important problem in academia.In this paper,we divided this paper into two parts based on transcoding and evaluation of encoders.The evaluation is the foreshadowing of the transcoding,we adopt different presets and encoders and test all the results based on 4K sequences.And to accelerate the transcoding speed,much effort were done on analysing the proportion of time in each module of x265 encoder.Based on this result,the second part mainly focus on the transcoding from H.264/AVC to HEVC.We extract important features and use SVM to train a mapping function.The x265 SKIPs recursive CU partition and reduces the amount of computation.The experimental results compared with x265 show that proposed alogrithm can effectively reduce 41.6% encoding time on average with about 2.79% increase in BD-Rate...
Keywords/Search Tags:transcoding, evaluation of encoders, SVM, faster CU splitting decison
Related items