基于改进的TLD目标跟踪算法
|
摘要
针对传统跟踪—学习—检测(tracking-learning-detecting,TLD)目标跟踪算法由于检测模块扫描大量子窗口而导致检测时间过长,并且在跟踪过程中当目标发生严重遮挡、形变时,TLD算法会出现跟踪失败的问题进行了研究,提出改进TLD目标跟踪算法。改进算法在检测模块前加入Vi Be模型预估前景目标,极大地缩小了检测区域。追踪模块用SIFT特征匹配算法来代替原算法中的光流法,准确跟踪目标避免发生跟踪漂移,减少了计算的复杂度,提高了算法适应环境的能力。实验表明,改进后的TLD算法运行速度得到提升,并且当目标出现严重遮挡、光照强度剧烈变化时的跟踪精度也得到了很好的改善。
Aiming at the problems of that the detecting module scans a large number of sub windows,which results the detection time is too long,and when the target has serious occlusion and deformation during the tracking process,the traditional tracking learning detection(TLD) target tracking algorithm will fail to track,so this paper proposed the improved TLD target tracking algorithm. Before the detection module,it added the ViBe model to estimate the foreground target,which greatly reduced the detection area. The tracking module used the SIFT feature matching algorithm to replace the optical flow method in the original algorithm,accurately tracked the target to avoid the tracking drift,reduced the complexity of the calculation and improved the ability of the algorithm to adapt to the environment. The experiment results show that the improved TLD algorithm can improve the running speed,and the tracking accuracy can also be improved when the target is seriously occluded and the light intensity changes dramatically.
Aiming at the problems of that the detecting module scans a large number of sub windows,which results the detection time is too long,and when the target has serious occlusion and deformation during the tracking process,the traditional tracking learning detection(TLD) target tracking algorithm will fail to track,so this paper proposed the improved TLD target tracking algorithm. Before the detection module,it added the ViBe model to estimate the foreground target,which greatly reduced the detection area. The tracking module used the SIFT feature matching algorithm to replace the optical flow method in the original algorithm,accurately tracked the target to avoid the tracking drift,reduced the complexity of the calculation and improved the ability of the algorithm to adapt to the environment. The experiment results show that the improved TLD algorithm can improve the running speed,and the tracking accuracy can also be improved when the target is seriously occluded and the light intensity changes dramatically.
引文
[1]童源,费树岷,沈捷.基于TLD框架的快速目标跟踪方法[J].计算机应用研究,2018,35(1):317-320.(Tong Yuan,Fei Shumin,Shen Jie.Fast target tracking method based on TLD framework[J].Application Research of Computers,2018,35(1):317-320.)
[2]郭秋滟.基于改进TLD算法的视频目标跟踪[J].计算机工程与设计,2017,38(9):2551-2555.(Guo Qiuyan.Video target tracking based on improved TLD algorithm[J].Computer Engineering and Design,2017,38(9):2551-2555.)
[3]周鑫,钱秋朦,叶永强,等.改进后的TLD视频目标跟踪方法[J].中国图象图形学报,2013,18(9):1115-1123.(Zhou Xin,Qian Qiumeng,Ye Yongqiang,et al.Improved TLD video object tracking method[J].Chinese Journal of Image and Graphics,2013,18(9):1115-1123.)
[4]王铁东,任世卿.一种改进TLD的目标跟踪算法[J].江苏科技信息,2018,35(1):52-53,56.(Wang Tiedong,Ren Shiqing.An improved TLD target tracking algorithm[J].Jiangsu Science and Technology Information,2008,35(1):52-53,56.)
[5]周军娜,陈伟,王珂,等.基于TLD的稀疏原型目标跟踪算法[J].计算机工程,2017,43(6):236-240.(Zhou Junna,Chen Wei,Wang Ke,et al.Sparse prototype object tracking algorithm based on TLD[J].Computer Engineering,2017,43(6):236-240.)
[6]刘曙,狄红卫,姚曼虹.基于自适应尺度的TLD目标跟踪算法[J].光学技术,2017,43(6):542-546.(Liu Shu,Di Hongwei,Yao Manhong.TLD target tracking algorithm based on adaptive scale[J].Optical Technology,2017,43(6):542-546.)
[7]Kalal Z,Mikolajczyk K,Matas J.Tracking-learning-detection[J].IEEE Trans on Pattern Analysis and Machine Intelligence,2012,34(7):1409-1422.
[8]杨玉锋,李伟彤,许磊,等.一种改进的TLD跟踪算法[J].科技创新与应用,2016(28):63-64.(Yang Yufeng,Li Weitong,Xu Lei,et al.An improved TLD tracking algorithm[J].Technological Innovation and Application,2016(28):63-64.)
[9]曲海成,单晓晨,孟煜,等.检测区域动态调整的TLD目标跟踪算法[J].计算机应用,2015,35(10):2985-2989.(Qu Haicheng,Shan Xiaochen,Meng Yu,et al.TLD target tracking algorithm for detecting regional dynamic adjustment[J].Computer Application,2015,35(10):2985-2989.)
[10]王振昊.基于TLD改进的人脸检测跟踪算法[J].科技创新导报,2013(22):50-51.(Wang Zhenhao.Improved face detection and tracking algorithm based on TLD[J].Science and Technology Innovation Guide,2013(22):50-51.)
[11]付苗,邢藏菊.帧间差法对TLD跟踪算法的改进[J].电子设计工程,2017,25(7):183-186.(Fu Miao,Xing Zangju.Improvement of TLD tracking algorithm by frame difference method[J].Electronic Design Engineering,2017,25(7):183-186.)
[12]龚小彪.基于TLD框架的目标跟踪算法研究[D].成都:西南交通大学,2014.(Gong Xiaobiao.Research on target tracking algorithm based on TLD framework[D].Chengdu:Southwest Jiaotong University,2014.)
[13]张丹,陈兴文,赵姝颖.基于改进的随机森林TLD目标跟踪方法[J].大连民族大学学报,2016,18(3):255-259.(Zhang Dan,Chen Xingwen,Zhao Shuying.Improved random forest TLD target tracking method[J].Journal of Dalian University of Nationalities,2016,18(3):255-259.)
[14]徐勇.基于TLD框架的目标跟踪算法[D].南京:南京航空航天大学,2017.(Xu Yong.Target tracking algorithm based on TLD framework[D].Nanjing:Nanjing University of Aeronautics and Astronautics,2017.)
[15]刘春,翟志强.改进的Vi Be运动目标检测算法[J].传感器与微系统,2017,36(1):123-126.(Liu Chun,Zhai Zhiqiang.Improved Vi Be motion target detection algorithm[J].Sensor and Microsystem,2017,36(1):123-126.)
[16]冯艳.动态背景下基于特征匹配的目标检测算法[D].西安:西安电子科技大学,2014.(Feng Yan.Target detection algorithm based on feature matching in dynamic context[D].Xi’an:Xi’an University of Electronic Science&Technology,2014.)
[17]傅卫平,秦川,刘佳,等.基于SIFT算法的图像目标匹配与定位[J].仪器仪表学报,2011,32(1):163-169.(Fu Weiping,Qin Chuan,Liu Jia,et al.Image target matching and location based on SIFT algorithm[J].Instrument Report,2011,32(1):163-169.)
[2]郭秋滟.基于改进TLD算法的视频目标跟踪[J].计算机工程与设计,2017,38(9):2551-2555.(Guo Qiuyan.Video target tracking based on improved TLD algorithm[J].Computer Engineering and Design,2017,38(9):2551-2555.)
[3]周鑫,钱秋朦,叶永强,等.改进后的TLD视频目标跟踪方法[J].中国图象图形学报,2013,18(9):1115-1123.(Zhou Xin,Qian Qiumeng,Ye Yongqiang,et al.Improved TLD video object tracking method[J].Chinese Journal of Image and Graphics,2013,18(9):1115-1123.)
[4]王铁东,任世卿.一种改进TLD的目标跟踪算法[J].江苏科技信息,2018,35(1):52-53,56.(Wang Tiedong,Ren Shiqing.An improved TLD target tracking algorithm[J].Jiangsu Science and Technology Information,2008,35(1):52-53,56.)
[5]周军娜,陈伟,王珂,等.基于TLD的稀疏原型目标跟踪算法[J].计算机工程,2017,43(6):236-240.(Zhou Junna,Chen Wei,Wang Ke,et al.Sparse prototype object tracking algorithm based on TLD[J].Computer Engineering,2017,43(6):236-240.)
[6]刘曙,狄红卫,姚曼虹.基于自适应尺度的TLD目标跟踪算法[J].光学技术,2017,43(6):542-546.(Liu Shu,Di Hongwei,Yao Manhong.TLD target tracking algorithm based on adaptive scale[J].Optical Technology,2017,43(6):542-546.)
[7]Kalal Z,Mikolajczyk K,Matas J.Tracking-learning-detection[J].IEEE Trans on Pattern Analysis and Machine Intelligence,2012,34(7):1409-1422.
[8]杨玉锋,李伟彤,许磊,等.一种改进的TLD跟踪算法[J].科技创新与应用,2016(28):63-64.(Yang Yufeng,Li Weitong,Xu Lei,et al.An improved TLD tracking algorithm[J].Technological Innovation and Application,2016(28):63-64.)
[9]曲海成,单晓晨,孟煜,等.检测区域动态调整的TLD目标跟踪算法[J].计算机应用,2015,35(10):2985-2989.(Qu Haicheng,Shan Xiaochen,Meng Yu,et al.TLD target tracking algorithm for detecting regional dynamic adjustment[J].Computer Application,2015,35(10):2985-2989.)
[10]王振昊.基于TLD改进的人脸检测跟踪算法[J].科技创新导报,2013(22):50-51.(Wang Zhenhao.Improved face detection and tracking algorithm based on TLD[J].Science and Technology Innovation Guide,2013(22):50-51.)
[11]付苗,邢藏菊.帧间差法对TLD跟踪算法的改进[J].电子设计工程,2017,25(7):183-186.(Fu Miao,Xing Zangju.Improvement of TLD tracking algorithm by frame difference method[J].Electronic Design Engineering,2017,25(7):183-186.)
[12]龚小彪.基于TLD框架的目标跟踪算法研究[D].成都:西南交通大学,2014.(Gong Xiaobiao.Research on target tracking algorithm based on TLD framework[D].Chengdu:Southwest Jiaotong University,2014.)
[13]张丹,陈兴文,赵姝颖.基于改进的随机森林TLD目标跟踪方法[J].大连民族大学学报,2016,18(3):255-259.(Zhang Dan,Chen Xingwen,Zhao Shuying.Improved random forest TLD target tracking method[J].Journal of Dalian University of Nationalities,2016,18(3):255-259.)
[14]徐勇.基于TLD框架的目标跟踪算法[D].南京:南京航空航天大学,2017.(Xu Yong.Target tracking algorithm based on TLD framework[D].Nanjing:Nanjing University of Aeronautics and Astronautics,2017.)
[15]刘春,翟志强.改进的Vi Be运动目标检测算法[J].传感器与微系统,2017,36(1):123-126.(Liu Chun,Zhai Zhiqiang.Improved Vi Be motion target detection algorithm[J].Sensor and Microsystem,2017,36(1):123-126.)
[16]冯艳.动态背景下基于特征匹配的目标检测算法[D].西安:西安电子科技大学,2014.(Feng Yan.Target detection algorithm based on feature matching in dynamic context[D].Xi’an:Xi’an University of Electronic Science&Technology,2014.)
[17]傅卫平,秦川,刘佳,等.基于SIFT算法的图像目标匹配与定位[J].仪器仪表学报,2011,32(1):163-169.(Fu Weiping,Qin Chuan,Liu Jia,et al.Image target matching and location based on SIFT algorithm[J].Instrument Report,2011,32(1):163-169.)