像素特征与粘连人体分割结合的人数统计方法
|
摘要
针对类似电影院、教室之类的人体间遮挡较少的场景,提出一种像素特征与粘连人体分割相结合的人数统计方法,为更好地处理粘连人体的分割问题,提出归一化距离度量的聚类算法与基于动态掩膜的投影法。当区域内人员较少时,通过建立区域内归一化后像素数与区域人数间的对应关系实现间接人数统计;当区域内人数增多且高于一定程度时,借助简单场景中人员特定的位置信息,进行粘连人体分割并对分割结果进行计数统计。在处理粘连人体分割问题上,针对人体连通区域的不同特点,分别使用归一化距离度量的聚类算法和改进的投影法进行人体粘连区域的行列分割。通过在影院影厅进行实验验证了该算法的有效性。
A method based on the method combining pixel and touched human object segmentation for less shielded scenes like cinemas and classrooms were put forward.To better deal with the segmentation problem of the adherent human body,a clustering algorithm based on the normalized distance measurement and the projection method based on the dynamic mask were proposed.When the number of people in the region was small,an indirect people counting method was used by establishing the corresponding relationship between the number of pixels and the number of regions after the normalization of the region.When the number of people in the region increased and was higher than a certain degree,the human body was segmented and the segmentation results were counted with the help of the specific location information in the simple scene.In dealing with the segmentation problem of adherent human body,aiming at the different characteristics of the connected area of human body,normalized distance metric clustering algorithm and improved projection method were used to do the segmentation of human adhesion regions.Results of expriments in the Theater Hall show that the proposed algorithm is effective.
A method based on the method combining pixel and touched human object segmentation for less shielded scenes like cinemas and classrooms were put forward.To better deal with the segmentation problem of the adherent human body,a clustering algorithm based on the normalized distance measurement and the projection method based on the dynamic mask were proposed.When the number of people in the region was small,an indirect people counting method was used by establishing the corresponding relationship between the number of pixels and the number of regions after the normalization of the region.When the number of people in the region increased and was higher than a certain degree,the human body was segmented and the segmentation results were counted with the help of the specific location information in the simple scene.In dealing with the segmentation problem of adherent human body,aiming at the different characteristics of the connected area of human body,normalized distance metric clustering algorithm and improved projection method were used to do the segmentation of human adhesion regions.Results of expriments in the Theater Hall show that the proposed algorithm is effective.
引文
[1]Tan Min,Hu Zhenfeng,Wang Baoyuan,et al.Robust object recognition via weakly supervised metric and template learning[J].Neurocomputing,2015,181:96-107.
[2]SHI Zenglin,YE Yangdong, WU Yunpeng,et al.Crowd counting using rank-based spatial pyramid pooling network[J].Acta Automatica Sinica,2016,42(6):866-874(in Chinese).[时增林,叶阳东,吴云鹏,等.基于序的空间金字塔池化网络的人群计数方法[J].自动化学报2016,42(6):866-873.]
[3]YE Feng,HONG Siting,ZHENG Decheng,et al.The people counting method based on Adaboost and background subtraction in indoor environment[J].Journal of Fujian Normal University(Natural Science Edition),2017,33(1):7-13(in Chinese).[叶锋,洪斯婷,郑德城,等.基于Adaboost与背景差分级联的室内人数统计方法[J].福建师范大学学报(自然科学版),2017,33(1):7-13.]
[4]Rao AS,Gubbi J,Marusic S,et al.Estimation of crowd density by clustering motion cues[J].The Visual Computer,2015:31(11):1533-1552.
[5]QIN Xunhui,WANG Xiufei,ZHOU Xi,et al.Counting people in various crowed density scenes using support vector regression[J].Journal of Image and Graphics,2013,18(4):392-398(in Chinese).[覃勋辉,王修飞,周曦,等.多种人群密度场景下的人群计数[J].中国图象图形学报,2013,18(4):392-398.]
[6]ZHANG Wenda,XU Yuelei,NI Jiacheng,et al.Image target recognition method based on multi-scale block convolution neural network[J].Journal of Computer Application,2016,36(4):1033-1038(in Chinese).[张文达,许悦雷,倪嘉成,等.基于多尺度分块卷积神经网络的图像目标识别算法[J].计算机应用,2016,36(4):1033-1038.]
[7]PENG Tianqiang,LI Fang.Image retrieval based on deep convolutional neural networks and binary hashing learning[J].Journal of Eleotronies&Information Technology,2016,38(8):2068-2075.
[8]HUANG Wenming,WEI Peng,LIANG Jinhua.Application of CNN-based hashing in image retrieval[J].Computer Engineering and Design,2017,38(2):517-521(in Chinese).[黄文明,魏鹏,梁金华.基金卷积神经网络的哈希在图像检索中的应用[J].计算机工程与设计,2017,38(2):517-521.]
[9]DING Yi,CHEN Shuyue,LIU Jinxing,et al.Crowd density estimation based on normalized target pixels[J].Computer Application and Software,2016,33(4):212-214(in Chinese).[丁艺,陈树越,刘金星,等.基于归一化目标像素的人群密度估计方法[J].计算机应用与软件,2016,33(4):212-214.]
[10]Liang Ronghua,Zhu Yuge, Wang Haixia.Counting crowd flow based on feature points[J].Neurocomputing,2014,133(8):377-384.
[11]LI Yang,ZHANG Yafei, MIAO Zhuang,et al.Convolutional encoding image retrieval method based on t-SNE[J].Application ResearchofComputers, 2017, 34(4):1244-1248.
[12]Conte D,Foggia P,Percannella G,et al.Counting moving persons in crowded scenes[J].Machine vision and Applications,2013,24(5):1029-1042.
[13]Liang R,Zhu Y,Wang H.Counting crowd flow based on feature points[J]. Neurocomputing,2014,133(8):377-384.
[14]Ryan D,Denman S,Sridharan S,et al.An evalution of crowd counting methods,features and regression models[J].Computer Vision and Image Understanding,2015,130(C):1-17.
[15]ZHANG Jingyang,PAN Guangyou.Comparison and application of multiple linear regression and BP neural network prediction model[J].Journal of Kunming University of Science and Technology(Natural Science Edition),2013,38(6):61-67(in Chinese).[张景阳,潘光有.多元线性回归与BP神经网络预测模型对比运用研究[J].昆明理工大学学报(自然科学版),2013,38(6):61-67.]
[16]LYU Jimin,ZENG Zhaoxian,ZHANG Maojun.Counting pedestrains in video sequences based on non-maxima suppression clustering[J].PR&AI,2012,25(1):150-155(in Chinese).[吕济民,曾昭贤,张茂军.基于非最大值抑制聚类的视频人数估测方法[J].模式识别与人工智能,2012,25(1):150-155.]
[2]SHI Zenglin,YE Yangdong, WU Yunpeng,et al.Crowd counting using rank-based spatial pyramid pooling network[J].Acta Automatica Sinica,2016,42(6):866-874(in Chinese).[时增林,叶阳东,吴云鹏,等.基于序的空间金字塔池化网络的人群计数方法[J].自动化学报2016,42(6):866-873.]
[3]YE Feng,HONG Siting,ZHENG Decheng,et al.The people counting method based on Adaboost and background subtraction in indoor environment[J].Journal of Fujian Normal University(Natural Science Edition),2017,33(1):7-13(in Chinese).[叶锋,洪斯婷,郑德城,等.基于Adaboost与背景差分级联的室内人数统计方法[J].福建师范大学学报(自然科学版),2017,33(1):7-13.]
[4]Rao AS,Gubbi J,Marusic S,et al.Estimation of crowd density by clustering motion cues[J].The Visual Computer,2015:31(11):1533-1552.
[5]QIN Xunhui,WANG Xiufei,ZHOU Xi,et al.Counting people in various crowed density scenes using support vector regression[J].Journal of Image and Graphics,2013,18(4):392-398(in Chinese).[覃勋辉,王修飞,周曦,等.多种人群密度场景下的人群计数[J].中国图象图形学报,2013,18(4):392-398.]
[6]ZHANG Wenda,XU Yuelei,NI Jiacheng,et al.Image target recognition method based on multi-scale block convolution neural network[J].Journal of Computer Application,2016,36(4):1033-1038(in Chinese).[张文达,许悦雷,倪嘉成,等.基于多尺度分块卷积神经网络的图像目标识别算法[J].计算机应用,2016,36(4):1033-1038.]
[7]PENG Tianqiang,LI Fang.Image retrieval based on deep convolutional neural networks and binary hashing learning[J].Journal of Eleotronies&Information Technology,2016,38(8):2068-2075.
[8]HUANG Wenming,WEI Peng,LIANG Jinhua.Application of CNN-based hashing in image retrieval[J].Computer Engineering and Design,2017,38(2):517-521(in Chinese).[黄文明,魏鹏,梁金华.基金卷积神经网络的哈希在图像检索中的应用[J].计算机工程与设计,2017,38(2):517-521.]
[9]DING Yi,CHEN Shuyue,LIU Jinxing,et al.Crowd density estimation based on normalized target pixels[J].Computer Application and Software,2016,33(4):212-214(in Chinese).[丁艺,陈树越,刘金星,等.基于归一化目标像素的人群密度估计方法[J].计算机应用与软件,2016,33(4):212-214.]
[10]Liang Ronghua,Zhu Yuge, Wang Haixia.Counting crowd flow based on feature points[J].Neurocomputing,2014,133(8):377-384.
[11]LI Yang,ZHANG Yafei, MIAO Zhuang,et al.Convolutional encoding image retrieval method based on t-SNE[J].Application ResearchofComputers, 2017, 34(4):1244-1248.
[12]Conte D,Foggia P,Percannella G,et al.Counting moving persons in crowded scenes[J].Machine vision and Applications,2013,24(5):1029-1042.
[13]Liang R,Zhu Y,Wang H.Counting crowd flow based on feature points[J]. Neurocomputing,2014,133(8):377-384.
[14]Ryan D,Denman S,Sridharan S,et al.An evalution of crowd counting methods,features and regression models[J].Computer Vision and Image Understanding,2015,130(C):1-17.
[15]ZHANG Jingyang,PAN Guangyou.Comparison and application of multiple linear regression and BP neural network prediction model[J].Journal of Kunming University of Science and Technology(Natural Science Edition),2013,38(6):61-67(in Chinese).[张景阳,潘光有.多元线性回归与BP神经网络预测模型对比运用研究[J].昆明理工大学学报(自然科学版),2013,38(6):61-67.]
[16]LYU Jimin,ZENG Zhaoxian,ZHANG Maojun.Counting pedestrains in video sequences based on non-maxima suppression clustering[J].PR&AI,2012,25(1):150-155(in Chinese).[吕济民,曾昭贤,张茂军.基于非最大值抑制聚类的视频人数估测方法[J].模式识别与人工智能,2012,25(1):150-155.]