基于稀疏表示的船体检测方法研究
|
摘要
为进一步强化航道安全,解决海事CCTV人工值守、非自动化问题,提出了基于稀疏表示的船体检测方法。利用稀疏表示实现对船体的检测时,首先构建样本特征矩阵,然后利用K-SVD算法对样本特征矩阵进行学习,得到冗余字典,最后对测试样本进行重构,根据马氏距离判断测试样本属性。通过与传统方法的试验比较,实验结果表明,该算法实时性好、检测准确率高,可以很好地对CCTV视频监控的船体进行检测与跟踪,解决CCTV人工值守、非自动化问题,节省大量人力资源。
In order to enhance securities of water-way and improve the effectiveness of the CCTV system, an approach based on sparse representation is proposed to automatically detect and track the vessels on the water-way. First, the dictionary of the samples is constructed by using the K-SVD algorithm to learn and train the samples' dictionary, and then the over-complete dictionary that can be used to represent the test samples is obtained. Finally, the mahalanobis distance between the test sample and the reconstructed sample is used to classify the test sample. This method is compared with the traditional methods.The experimental results show that the effectiveness of the vessel detection based on SR outperforms the traditional SVM method in the efficiency and the accuracy, which can solve the vessel detection problem.
In order to enhance securities of water-way and improve the effectiveness of the CCTV system, an approach based on sparse representation is proposed to automatically detect and track the vessels on the water-way. First, the dictionary of the samples is constructed by using the K-SVD algorithm to learn and train the samples' dictionary, and then the over-complete dictionary that can be used to represent the test samples is obtained. Finally, the mahalanobis distance between the test sample and the reconstructed sample is used to classify the test sample. This method is compared with the traditional methods.The experimental results show that the effectiveness of the vessel detection based on SR outperforms the traditional SVM method in the efficiency and the accuracy, which can solve the vessel detection problem.
引文
[1]洪晶.海事场景的视频监控系统[D].浙江大学,2007.
[2]李斌,孙华峰.内河运动船舶的准确检测跟踪技术研究[J].微电子学与计算机,2012,29(11):180-184.LI Bin,SUN Hua-feng.Research the inland river motion ships accurately detection and tracking algorithm[J].Microelectronics,2012,29(11):180-184.
[3]叶玲利.内河视频监控中运动船舶检测和遮挡分离算法研究[D].汉理工大学,2013.
[4]Olshausen B A,Field J.Sparse coding with an over-complete basis set:a strategy employed by v1[J].Vision Research,1997,37(23):3311-3325.
[5]Wei Tang,Zhenwei Shi,Zhana Duren.Sparse hyperspectral unmixing using an approximate L0 norm[J].Optick,2013,125(2014):31-38.
[6]李修志,吴健,崔志明,等.复杂交通场景中采用稀疏表示的车辆识别方法[J].中国图像图形学报,2012,17(3):387-392.LI Xiu-zhi,WU Jian,CUI Zhi-ming,et al.Sparse representation method of vehicle recognition in complex traffic scenes[J].Journal of Image and Graphics,2012,17(3):387-392.
[7]Needell D,Tropp J.Co Sa MP:Iterative signal recovery from incomplete and inaccurate samples[J].Communications of the ACM,2010,53(12):93-100.
[8]肖玲,李仁发,曾凡仔,等.基于自学习稀疏表示的动态手势识别方法[J].通信学报,2013,34(6):128-135.XIAO Ling,LI Ren-fa,ZENG Fan-zai,et al.Gesture recognition approach based on learning sparse representation[J].Journal on Communications,2013,34(6):128-135.
[9]Pati Y C,Reza ifar R.K rishnaprased.Orthogonal matching pursuit:recursive function approximation with applications to wave le t decomposition[A]//Proceedings of the 27th Annual Asilomar Conference in Signals,Systems and Computers,November 1-3,1993[C].Pacific Grove:IEEE Explore,1993.
[10]尹忠科,邵君,Pierre Vandergheynst.利用FFT实现基于MP的信号稀疏分解[J].电子信息学报,2006,28(4):614-618.YIN Zhong-ke,SHAO Jun,Pierre Vandergheynst.MP Based Signal Sparse Decomposition with FFT[J].Journal of Electronics&Information Technology,2006,28(4):614-618.
[11]Dalal N,Triggs B.Histograms of Oriented Gradients forHuman Detection[M].CVPR,005.
[12]陈中杰,蒋刚,蔡勇.基于SVM一对一多分类算法的二次细分法研究[J].传感器与微系统,2013,32(4):44-47.CHEN Zhong-jie,JIANG Gang,CAI Yong.Research of secondary subdivision method for one-versus-one multi-classification algorithm based on SVM[J].Transducer and Microsystem Technologies,2013,32(4):44-47.
[2]李斌,孙华峰.内河运动船舶的准确检测跟踪技术研究[J].微电子学与计算机,2012,29(11):180-184.LI Bin,SUN Hua-feng.Research the inland river motion ships accurately detection and tracking algorithm[J].Microelectronics,2012,29(11):180-184.
[3]叶玲利.内河视频监控中运动船舶检测和遮挡分离算法研究[D].汉理工大学,2013.
[4]Olshausen B A,Field J.Sparse coding with an over-complete basis set:a strategy employed by v1[J].Vision Research,1997,37(23):3311-3325.
[5]Wei Tang,Zhenwei Shi,Zhana Duren.Sparse hyperspectral unmixing using an approximate L0 norm[J].Optick,2013,125(2014):31-38.
[6]李修志,吴健,崔志明,等.复杂交通场景中采用稀疏表示的车辆识别方法[J].中国图像图形学报,2012,17(3):387-392.LI Xiu-zhi,WU Jian,CUI Zhi-ming,et al.Sparse representation method of vehicle recognition in complex traffic scenes[J].Journal of Image and Graphics,2012,17(3):387-392.
[7]Needell D,Tropp J.Co Sa MP:Iterative signal recovery from incomplete and inaccurate samples[J].Communications of the ACM,2010,53(12):93-100.
[8]肖玲,李仁发,曾凡仔,等.基于自学习稀疏表示的动态手势识别方法[J].通信学报,2013,34(6):128-135.XIAO Ling,LI Ren-fa,ZENG Fan-zai,et al.Gesture recognition approach based on learning sparse representation[J].Journal on Communications,2013,34(6):128-135.
[9]Pati Y C,Reza ifar R.K rishnaprased.Orthogonal matching pursuit:recursive function approximation with applications to wave le t decomposition[A]//Proceedings of the 27th Annual Asilomar Conference in Signals,Systems and Computers,November 1-3,1993[C].Pacific Grove:IEEE Explore,1993.
[10]尹忠科,邵君,Pierre Vandergheynst.利用FFT实现基于MP的信号稀疏分解[J].电子信息学报,2006,28(4):614-618.YIN Zhong-ke,SHAO Jun,Pierre Vandergheynst.MP Based Signal Sparse Decomposition with FFT[J].Journal of Electronics&Information Technology,2006,28(4):614-618.
[11]Dalal N,Triggs B.Histograms of Oriented Gradients forHuman Detection[M].CVPR,005.
[12]陈中杰,蒋刚,蔡勇.基于SVM一对一多分类算法的二次细分法研究[J].传感器与微系统,2013,32(4):44-47.CHEN Zhong-jie,JIANG Gang,CAI Yong.Research of secondary subdivision method for one-versus-one multi-classification algorithm based on SVM[J].Transducer and Microsystem Technologies,2013,32(4):44-47.