实时视频去隔行的并行算法研究

南京大学学报(自然科学版) ›› 2016, Vol. 52 ›› Issue (5): 795–.

实时视频去隔行的并行算法研究

余双春，袁　杰，沈庆宏*

出版日期:2016-09-24 发布日期:2016-09-24
作者简介: 南京大学电子科学与工程学院，南京，210093
基金资助:
基金项目：国家自然科学基金(61201425)，江苏省自然科学基金(BK20131280)
收稿日期：2015－12－28
*通讯联系人，Email：qhshen@nju.edu.cn

Research on realtime video deinterlace with parallel computing

Yu Shuangchun，Yuan Jie，Sheng Qinghong*

Online:2016-09-24 Published:2016-09-24
About author: School of Electronic Science and Engineering，Nanjing University，Nanjing，210093，China

摘要/Abstract

摘要： 为了实现实时将隔行扫描信号转换为逐行扫描信号的目标，采用的是运动自适应去隔行方法，并配以多线程并行计算．运动自适应相对单场插值利用了图像在时间上的关联性，在运动区域、静止区域都能有更优的效果；相对于运动补偿，具有时间复杂度上的优越性，为实时处理做铺垫．运动自适应方法，采用了更好的运动检测方法，并通过运动修正更准确地划分动静区域，场内采用一种最大相关性的保持边缘方法进行插值，更好的保护图像的细节信息，此方法时间复杂度低，且在信噪比上优于其他常见运动自适应方法．基于多核服务器，进行多线程并行编程，充分利用了多核体系，计算效率大幅度提高．实验结果表明，方法处理后的图像效果相比其他方法更佳，实现了实时处理D1标准视频序列．

Abstract: In the television broadcasting system，considering for the limit of transmission bandwidth，television signals will continue to use interlaced format for a long time.Digital television only perfectly display progressive scanning video，and the original display of interlaced video on these digital display screen，could cause line crawl，parallel phenomenon，serration and line flicker and so on.Thus，deinterlacing is a necessary step before the TV image is passed to the display system.A novel realtime deinterlacing algorithm is proposed in this paper.Realtime deinterlacing algorithm is based on motion adaptive deinterlacing and multithread parallel computing.Deinterlacing algorithm contains intrafiled deinterlacing，interfield deinterlacing，and blending deinterlacing.Blending deinterlacing contains motion adaptive deinterlacing and motioncompensated method.The motion adaptive deinterlacing obtains the advantages of both interfield deinterlacing and intrafield deinterlacing.Compared with motioncompensated method，the motion adaptive deinterlacing costs resource more little and processes quicker.It detects the motion areas first，and in this paper，a better detecting method is used.Then adopts intrafield deinterlacing in motion areas，and in this paper，it takes a max correlation method to preserve image edges.Interfield deinterlacing adopt in static areas，this paper uses twofiled average method.To achieve the goal of realtime deinterlacing，multithread parallel computing is applied to deinterlacing.Nowadays，as the multicore processors becoming more and more popular，the development of parallel software technology is proved to be a crux.Widows multithreaded API，POSIX threads API、openMP and languages that support multithread programming，these are typical methods.This paper using the C and multithreaded API，makes computing efficient，makes it possible for realtime deinterlacing.The result of experiment indicates that the realtime deinterlacing algorithm proposed in this paper plays a very good effect.

余双春，袁　杰，沈庆宏*. 实时视频去隔行的并行算法研究[J]. 南京大学学报(自然科学版), 2016, 52(5): 795–.

Yu Shuangchun，Yuan Jie，Sheng Qinghong*. Research on realtime video deinterlace with parallel computing[J]. Journal of Nanjing University(Natural Sciences), 2016, 52(5): 795–.

参考文献

[1]　de Haan G，Bellers E B．Deinterlacingan overview.In Proceedings of the IEEE，1998，86(9)：1839－1857.
[2] 赵娜娜，王向文，刘顺兰等．基于中值滤波与边缘插值的视频去隔行算法．杭州电子科技大学学报，2011，31(5)：119－122.(Zhao N N，Wang X W，Liu S L，et al.Video deinterlacing algorithm based on median filtering and edgedependent interpolation algorithm.Journal of Hangzhou Dianzi University，2011，31(5)：119－122.)
[3] 聂　苗，黎　英，石力卓等．基于视频监控系统的运动自适应去隔行算法．计算机应用，2013，33(10)：2922－2925，2930.(Nie M，Li Y，Shi L Z，et al.Video surveillance systembased motionadaptive deinterlacing algorithm.Journal of Computer Applications，2013，33(10)：2922－2925，2930.)
[4] Lin S F，Chang Y L，Chen L G．Motion adaptive interpolation with horizontal motion detection for deinterlacing.IEEE Transactions on Consumer Electronics，2003，49(4)：1256－1265.
[5] Yoo H，Jeong J C．Directionoriented interpolation and its application to deinterlacing.IEEE Transactions on Consumer Electronics，2002，48(4)：954－962.
[6] Park M K，Kang M G，Nam K，et al.New edge dependent deinterlacing algorithm based on horizontal edge pattern.IEEE Transactions on Consumer Electronics，2003，49(4)：1508－1512.
[7] Yang R E，Che K，Li J X．A motion adaptive deinterlacing algorithm based on edgedependent interpolation.In：Control and Decision Conference(CCDC)，2010 Chinese，2010，910－913.
[8] Qiang J，Chen J，Wang J．An interfield and intrafield weighted interpolative deinterlacing algorithm based on lowangle detection and multiangle extraction.Mathematical Problems in Engineering，2014，2014(3)：235－254.
[9] 岳雷飞．视频格式转换中去隔行算法的研究．硕士学位论文．合肥：合肥工业大学，2010.(Yue L F．Research of deinterlacing algorithm of video format conversion.Master Dissertation.Hefei：Hefei University of Technology，2010.)
[10] 孙晨艳．基于运动补偿的去隔行算法研究．硕士学位论文．西安：西安电子科技大学，2013.(Sun Y C．Deinterlacing algorithm based on motion compensation.Master Dissertation.Xi’an：University of Electronic Science and Technology，2013.)
[11] 许卫全，陈建辉，徐中佑等．一种具有高精度运动检测的自适应运动补偿去隔行算法．信号处理，2006，22(5)：658－662.(Xu W Q，Chen J H，Xu Z Y，et al.A new motion adaptive deinterlace algorithm with high precise motion detection.Signal Processing，2006，22(5)：658－662.)
[12] Kim W K，Jin S J，Jeong J C．Novel intra deinterlacing algorithm using content adaptive interpolation.IEEE Transactions on Consumer Electronics，2007，53(3)：1036－1043.
[13] 马　斌，丁　勇，周　升等．加权边沿自适应的场内插值去隔行方法．计算机应用研究，2010，27(8)：3168－3170，3175.(Ma B，Ding Y，Zhou S，et al.Deinterlacing with weighted edge adaptive intrafield interpolation.Application Research of Computers，2010，27(8)：3168－3170，3175.)
[14] Yao Y Y．Stratified rough sets and granular computing.In：Dave R N，Sudkamp T．In：The 18th International Conference of the North American Fuzzy Information Processing Society.New York：IEEE Press，1999，800－804.
[15] 冯　伟，袁　杰．高清视频并行处理的研究．南京大学学报(自然科学)，2012，48(1)：33－39.(Feng W，Yuan J．A study on parallel processing for highdefinition video.Journal of Nanjing University(Natural Sciences)，2012，48(1)：33－39.)
[16] Wu J J，Song Z，Jeon G．GPUparallel implementation of the edgedirected adaptive intrafield deinterlacing method.Display Technology Journal，2014，10(9)：746－753.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed