基于多尺度感知哈希特征的目标跟踪算法研究
|
摘要
研究了目标跟踪与检测。考虑到其应用环境复杂,造成对光照变化鲁棒性不强,目标跟踪过程中数据丢失,实时跟踪时检测速度不够的问题,结合哈希指纹特征对光照变化鲁棒性强的特点,提出了一种基于多尺度(multi-scale)感知哈希(Hash)特征的目标跟踪算法,即MHash算法。该算法采用多种尺度扫描窗口对待检测图像进行扫描,利用双线性内插值的方法缩小扫描窗口提取哈希指纹特征;同时利用汉明距离与高斯分布相结合的方法对扫描窗口进行评价,评分最高的作为目标窗口。该算法已应用于人脸、运动车辆及快速和缓慢移动物体的跟踪与检测,并分析了跟踪与检测结果。结果表明,MHash算法具有速度快、跟踪稳定及对光照、目标尺度变化鲁棒性强的特点,能够长期有效地对目标进行跟踪。
Target tracking and detection are studied. Considering that the complex application environments bring the problems of low robustness to illumination variation,data loss due to the change of target scale,and low detecting speed during real-time tracking,a new object tracking algorithm using multi-scale perceptual Hash feature,called the MHash algorithm for short,is proposed based on the characteristic that Hash fingerprint features are robust to illumination changes. The algorithm uses multi-scale scanning windows to detect objects and reduces the dimension of scanning window to access perceptual Hash fingerprint characteristics with the bilinear interpolation method. And it combines the Hamming distance with the Gaussian distribution method to evaluate scanning windows. The window of the highest score is as the target window. The method is applied to tracking and detection of face,sports cars and fast and slow moving objects,and the experiments are analysed. The experiments show that this algorithm is robust against illumination changes and target scale change. It can be effectively used for long-term target tracking.
Target tracking and detection are studied. Considering that the complex application environments bring the problems of low robustness to illumination variation,data loss due to the change of target scale,and low detecting speed during real-time tracking,a new object tracking algorithm using multi-scale perceptual Hash feature,called the MHash algorithm for short,is proposed based on the characteristic that Hash fingerprint features are robust to illumination changes. The algorithm uses multi-scale scanning windows to detect objects and reduces the dimension of scanning window to access perceptual Hash fingerprint characteristics with the bilinear interpolation method. And it combines the Hamming distance with the Gaussian distribution method to evaluate scanning windows. The window of the highest score is as the target window. The method is applied to tracking and detection of face,sports cars and fast and slow moving objects,and the experiments are analysed. The experiments show that this algorithm is robust against illumination changes and target scale change. It can be effectively used for long-term target tracking.
引文
[1]Chen L,Wei H,Ferryman J.A survey of human motion analysis using depth imagery[J].Pattern Recognition Letters,2013,34(15):1995-2006
[2]Lara O D,Labrador M A.A Survey on human activity recognition using wearable sensors[J].IEEE Communications Surveys&Tutorials,2013,15(3):1192-1209
[3]Jiang Q.Research of multiple-instance learning for target recognition and tracking[J].Eurasip Journal on Embedded Systems,2016,2016(1):1-6
[4]Wang S,Fowlkes C C.Learning optimal parameters for multi-target tracking with contextual interactions[J].International Journal of Computer Vision,2016,122:1-18
[5]Rautaray S S,Agrawal A.Vision based hand gesture recognition for human computer interaction:a survey[J].Artificial Intelligence Review,2015,43(1):1-54
[6]Li K,He F Z,Chen X,et al.Real-time object tracking via compressive feature selection[J].Frontiers of Computer Science,2016,10(4):689-701
[7]付巍,郑宾.基于改进的粒子滤波的静电目标跟踪算法[J].高技术通讯,2014,24(2):138-143
[8]邵万开,石澄贤.动态背景下运动目标检测与跟踪研究[J].计算机测量与控制,2016,24(8):52-55
[9]蓝龙,张翔,骆志刚.视觉目标跟踪现状与发展[J].国防科技,2017,38(05):12-18
[10]修春波,魏世安,万蓉凤.二维联合特征模型的自适应均值漂移目标跟踪[J].光电子·激光,2015(2):342-351
[11]郇二洋,李睿.基于自适应特征融合的粒子滤波目标跟踪算法[J].计算机科学,2015,42(2):316-318
[12]Zhang W Z,Ji J G,Jing Z Z,et al.Adaptive Real-Time Compressive Tracking[C].In:Proceedings of the International Conference on Network and Information Systems for Computers,Wuhan,China,2016:236-240
[13]Kalal Z,Mikolajczyk K,Matas J.Tracking-learning-detection[J].IEEE Transactions on Pattern Analysis&Machine Intelligence,2012,34(7):1409
[14]汪鸿翔,柳培忠,骆炎民,等.单目标跟踪算法综述[J].海峡科学,2016(7):28-32
[15]罗海波,许凌云,惠斌,等.基于深度学习的目标跟踪方法研究现状与展望[J].红外与激光工程,2017,46(5):6-12
[16]Dicle C,Camps O I,Sznaier M.The way they move:tracking multiple targets with similar appearance[C].In:Proceedings of the IEEE International Conference on Computer Vision,Sydney,Australia,2014.2304-2311
[17]Zhou H,Yuan Y,Shi C.Object tracking using SIFT features and mean shift[J].Computer Vision&Image Understanding,2009,113(3):345-352
[18]Hadmi A,Puech W,Said B A E,et al.A robust and secure perceptual hashing system based on a quantization step analysis[J].Signal Processing Image Communication,2013,28(8):929-948
[19]Pei B,Wang S Z,Li-Jia N I.Content-based image retrieval using perceptual image Hashing[J].Journal of Shanghai University,2012,18(4):335-341
[20]Yu S,Jiang Z.Visual tracking via perceptual image hash from a mobile robot[C].In:Proceedings of IEEE International Conference on Information and Automation,Lijiang,China,2015.1612-1616
[21]Zhang Y,Zhu C,Bres S,et al.Encoding local binary descriptors by bag-of-features with hamming distance for visual object categorization[C].In:Proceedings of the European Conference on Advances in Information Retrieval,Moscow,Russia,2016.630-641
[22]张庆刚,陈玮.融合感知哈希特征的时空上下文目标跟踪算法[J].信息技术,2017(4):144-147
[2]Lara O D,Labrador M A.A Survey on human activity recognition using wearable sensors[J].IEEE Communications Surveys&Tutorials,2013,15(3):1192-1209
[3]Jiang Q.Research of multiple-instance learning for target recognition and tracking[J].Eurasip Journal on Embedded Systems,2016,2016(1):1-6
[4]Wang S,Fowlkes C C.Learning optimal parameters for multi-target tracking with contextual interactions[J].International Journal of Computer Vision,2016,122:1-18
[5]Rautaray S S,Agrawal A.Vision based hand gesture recognition for human computer interaction:a survey[J].Artificial Intelligence Review,2015,43(1):1-54
[6]Li K,He F Z,Chen X,et al.Real-time object tracking via compressive feature selection[J].Frontiers of Computer Science,2016,10(4):689-701
[7]付巍,郑宾.基于改进的粒子滤波的静电目标跟踪算法[J].高技术通讯,2014,24(2):138-143
[8]邵万开,石澄贤.动态背景下运动目标检测与跟踪研究[J].计算机测量与控制,2016,24(8):52-55
[9]蓝龙,张翔,骆志刚.视觉目标跟踪现状与发展[J].国防科技,2017,38(05):12-18
[10]修春波,魏世安,万蓉凤.二维联合特征模型的自适应均值漂移目标跟踪[J].光电子·激光,2015(2):342-351
[11]郇二洋,李睿.基于自适应特征融合的粒子滤波目标跟踪算法[J].计算机科学,2015,42(2):316-318
[12]Zhang W Z,Ji J G,Jing Z Z,et al.Adaptive Real-Time Compressive Tracking[C].In:Proceedings of the International Conference on Network and Information Systems for Computers,Wuhan,China,2016:236-240
[13]Kalal Z,Mikolajczyk K,Matas J.Tracking-learning-detection[J].IEEE Transactions on Pattern Analysis&Machine Intelligence,2012,34(7):1409
[14]汪鸿翔,柳培忠,骆炎民,等.单目标跟踪算法综述[J].海峡科学,2016(7):28-32
[15]罗海波,许凌云,惠斌,等.基于深度学习的目标跟踪方法研究现状与展望[J].红外与激光工程,2017,46(5):6-12
[16]Dicle C,Camps O I,Sznaier M.The way they move:tracking multiple targets with similar appearance[C].In:Proceedings of the IEEE International Conference on Computer Vision,Sydney,Australia,2014.2304-2311
[17]Zhou H,Yuan Y,Shi C.Object tracking using SIFT features and mean shift[J].Computer Vision&Image Understanding,2009,113(3):345-352
[18]Hadmi A,Puech W,Said B A E,et al.A robust and secure perceptual hashing system based on a quantization step analysis[J].Signal Processing Image Communication,2013,28(8):929-948
[19]Pei B,Wang S Z,Li-Jia N I.Content-based image retrieval using perceptual image Hashing[J].Journal of Shanghai University,2012,18(4):335-341
[20]Yu S,Jiang Z.Visual tracking via perceptual image hash from a mobile robot[C].In:Proceedings of IEEE International Conference on Information and Automation,Lijiang,China,2015.1612-1616
[21]Zhang Y,Zhu C,Bres S,et al.Encoding local binary descriptors by bag-of-features with hamming distance for visual object categorization[C].In:Proceedings of the European Conference on Advances in Information Retrieval,Moscow,Russia,2016.630-641
[22]张庆刚,陈玮.融合感知哈希特征的时空上下文目标跟踪算法[J].信息技术,2017(4):144-147