Research on personnel re-recognition method in coal mine underground based on improved metric learning
-
摘要: 传统基于度量学习的煤矿井下行人重识别方法中，由于度量学习忽略正负样本绝对距离，造成损失函数梯度消失或梯度弥散，导致井下人员位置信息识别精度不高。针对该问题，提出了一种基于改进度量学习的煤矿井下行人重识别方法。首先，采用基于手工设计特征的井下人员特征提取方法，对颜色空间、纹理空间等特征进行手动加工提炼，丰富特征维度。然后，采用欧氏距离对人员高维特征进行相似性计算。最后，提出一种改进的三重损失函数，通过在传统三重损失函数中加入自适应权重，增加有效样本的权重，解决了由于忽略正负样本绝对距离导致的梯度消失或梯度弥散问题。将传统识别方法与基于改进度量学习的煤矿井下行人重识别方法进行了累积匹配特征曲线验证、识别速率验证，结果表明：① 基于改进度量学习的煤矿井下行人重识别方法在相似样本个数为50左右时，样本匹配概率达100%。② 在2种不同标定大小图像的推理耗时上，基于改进度量学习的煤矿井下行人重识别方法较传统重识别方法分别减少了44，68 ms。③ 基于改进度量学习的煤矿井下行人重识别方法在舍弃行人头脚部分图像后表现更好，在相似样本个数为42左右时，样本匹配概率达100%。Abstract: In the traditional personnel re-recognition method in coal mine underground based on metric learning, because metric learning ignores the absolute distance between positive and negative samples, the gradient of the loss function disappears or disperses. This results in low recognition precision of underground personnel position information. In order to solve this problem, a personnel re-recognition method in coal mine underground based on improved metric learning is proposed. Firstly, a feature extraction method for underground personnel based on manual design features is adopted to manually process and extract features such as color space and texture space, enriching the feature dimensions. Secondly, Euclidean distance is used to calculate the similarity of high-dimensional features of personnel. Finally, an improved triple loss function is proposed. Adding adaptive weights to the traditional triple loss function increases the weight of effective samples. It solves the problem of gradient disappearance or dispersion caused by ignoring the absolute distance between positive and negative samples. The traditional recognition method is compared with the personnel re-recognition method in coal mine underground based on improved metric learning for cumulative matching feature curve verification and recognition rate verification. The results show the following points. ① The personnel re-recognition method in coal mine underground based on improved metric learning has a sample matching probability of 100% when the number of similar samples is around 50. ② The personnel re-recognition method in coal mine underground based on improved metric learning reduces the inference time of two different calibration size images by 44 ms and 68 ms, respectively, compared to traditional re-recognition methods. ③ The personnel re-recognition method in coal mine underground based on improved metric learning performs better after discarding the images of personnel heads and feet. It has a sample matching probability of 100% when the number of similar samples is around 42.
-
-
![]()
图 3 传统与自适应的三重损失函数的函数曲线
Figure 3. Function curves of traditional and an adaptive trip loss function
![]()
图 5 基于传统度量学习与改进度量学习的行人重识别效率
Figure 5. Rerecognition efficiency under traditional heavy recognition and adaptive metric learning
![]()
图 6 舍弃头脚部信息后得出的自适应的三重损失下重识别效率
Figure 6. Heavy identification efficiency of adaptive metric learning after discarding head and feet information
表 1 手工设计特征指标
Table 1 Manual design feature indicators
特征类型 特征指标 颜色空间 HSV(Hue, Saturation, Value)，RGB(Red, Green, Blue)，Lab(Lab color space) 纹理空间 LBP(Local Binary Pattern, 局部二值模式)，加博尔滤波器 局部特征 尺度不变特征转换，方向梯度直方图 专用特征 对称局部特征累加，局部特征的集成，局部最大重现特征、显著特征匹配 
下载: 导出CSV
表 2 传统与改进的度量学习的行人重识别推理耗时
Table 2 The time cost between traditional rerecognition and adaptive metric learning
推理方法 测试图大小 测试图张数 测试次数 推理平均耗时/ms 基于传统度量学习
的行人重识别方法224×224 600 1000 248 基于改进度量学习
的行人重识别方法224×224 600 204 基于传统度量学习
的行人重识别方法640×640 400 413 基于改进度量学习
的行人重识别方法640×640 400 345
下载: 导出CSV
-
[1] 王国法,赵国瑞,任怀伟. 智慧煤矿与智能化开采关键核心技术分析[J]. 煤炭学报,2019,44(1):34-41. WANG Guofa,ZHAO Guorui,REN Huaiwei. Analysis on key technologies of intelligent coal mine and intelligent mining[J]. Journal of China Coal Society,2019,44(1):34-41.
[2] 孙继平. 煤矿智能化与矿用5G[J]. 工矿自动化,2020,46(8):1-7. SUN Jiping. Coal mine intelligence and mine-used 5G[J]. Industry and Mine Automation,2020,46(8):1-7.
[3] 张华,李靖锋,魏红磊,等. 基于智能视频识别技术的智能化煤矿安全管理研究与应用[J]. 工矿自动化,2021,47(增刊1):10-13. ZHANG Hua,LI Jingfeng,WEI Honglei,et al. Research and application of intelligent coal mine safety management based on intelligent video recognition technology[J]. Industry and Mine Automation,2021,47(S1):10-13.
[4] 张立亚. 矿山智能视频分析与预警系统研究[J]. 工矿自动化,2017,43(11):16-20. ZHANG Liya. Research on intelligent video analysis and early warning system for mine[J]. Industry and Mine Automation,2017,43(11):16-20.
[5] GHEISSARI N, SEBASTIAN T. B, HARTLEY R. Person reidentification using spatiotemporal appearance[C]. IEEE Computer Society Conference on Computer Vision and Pattern Recognition （CVPR）, New York, 2006: 1528-1535.
[6] GRAY D, BRENNAN S, TAO H. Evaluating appearance models for recognition reacquisition and tracking[C]. IEEE International Workshop on Performance Evaluation for Tracking and Surveillance （PETS）, Rio de Janeiro, 2007: 123-127.
[7] 苏荻翔,王帮海,叶子成. 融合多聚类信息的无监督行人重识别算法[J]. 计算机工程与应用,2022,58(18):195-204. SU Dixiang,WANG Banghai,YE Zicheng. Unsupervised person re-identification algorithm by fusing multi-cluster information[J]. Computer Engineering and Applications,2022,58(18):195-204.
[8] YE Mang,LAN Xiangyuan,WANG Zheng,et al. Bi-directional center-constrained top-ranking for visible thermal person re-identification[J]. IEEE Transactions on Information Forensics and Security,2019,15:407-419.
[9] QIN Danfeng, GAMMETER S, BOSSARD L, et al. Hello neighbor: accurate object retrieval with k-reciprocal nearest neighbors[C]. CVPR 2011, Colorado Spings, 2011: 777-784.
[10] 周智恒,刘楷怡,黄俊楚,等. 一种基于等距度量学习策略的行人重识别改进算法[J]. 电子与信息学报,2019,41(2):477-483. ZHOU Zhiheng,LIU Kaiyi,HUANG Junchu,et al. Improved metric learning algorithm for person re-identification based on equidistance[J]. Journal of Electronics & Information Technology,2019,41(2):477-483.
[11] 章东平,徐丽园. 距离度量学习的摄像网络中行人重识别[J]. 中国计量大学学报,2016,27(4):424-428,434. ZHANG Dongping,XU Liyuan. Human re-identification over camera networks by using distance metric learning[J]. Journal of China University of Metrology,2016,27(4):424-428,434.
[12] 刘浩,刘海滨,孙宇,等. 煤矿井下员工不安全行为智能识别系统[J]. 煤炭学报,2021,46(增刊2):1159-1169. LIU Hao,LIU Haibin,SUN Yu,et al. Intelligent recognition system of unsafe behavior of underground coal miners[J]. Journal of China Coal Society,2021,46(S2):1159-1169.
[13] 苏伟根. 基于图度量学习的行人重识别算法研究[D]. 广州: 广东工业大学, 2021. SU Weigen. Research on graph metric learning fo person re-identification[D]. Guangzhou: Guangdong University of Technology, 2021.
[14] 董映宇. 基于特征融合与度量学习的行人重识别研究[D]. 马鞍山: 安徽工业大学, 2020. DONG Yingyu. Research on pedestrian re-identification in based on fusion feature and metric learning[D]. Ma'anshan: Anhui University of Technology, 2020.
[15] MCPHERSON M,SMITH-LOVIN L,COOK J M. Birds of a feather:homophily in social networks[J]. Annual Review of Sociology,2001,27(1):415-444. DOI: 10.1146/annurev.soc.27.1.415
[16] 陈伟,任鹏,田子建,等. 基于注意力机制的无监督矿井人员跟踪[J]. 煤炭学报,2021,46(增刊1):601-608. DOI: 10.13225/j.cnki.jccs.2020.1348 CHEN Wei,REN Peng,TIAN Zijian,et al. Unsupervised mine personnel tracking based on attention mechanism[J]. Journal of China Coal Society,2021,46(S1):601-608. DOI: 10.13225/j.cnki.jccs.2020.1348
[17] 宋婉茹,赵晴晴,陈昌红,等. 行人重识别研究综述[J]. 智能系统学报,2017,12(6):770-780. SONG Wanru,ZHAO Qingqing,CHEN Changhong,et al. Survey on pedestrian re-identification research[J]. CAAI Transactions on Intelligent Systems,2017,12(6):770-780.
[18] MIGNON A, JURIE F. PCCA: a new approach for distance learning from sparsepairwise constraints[C]. IEEE Conference on Computer Vision and Pattern Recognition, Providence, 2012: 2666-2672.
[19] 张立亚,孟庆勇,杨坤. 基于维纳滤波的矿井监控图像的复原技术[J]. 煤矿安全,2019,50(1):129-132. ZHANG Liya,MENG Qingyong,YANG Kun. Recovery technology of mine monitoring image based on wiener filtering[J]. Safety in Coal Mines,2019,50(1):129-132.
[20] 张立亚,郝博南,孟庆勇,等. 基于HSV空间改进融合Retinex算法的井下图像增强方法[J]. 煤炭学报,2020,45(增刊1):532-540. ZHANG Liya,HAO Bonan,MENG Qingyong,et al. Method of image enhancement in coal mine based on improved retex fusion algorithm in HSV space[J]. Journal of China Coal Society,2020,45(S1):532-540.
[21] 丁宗元. 基于度量学习的行人重识别若干算法研究[D]. 常州: 常州大学, 2018. DING Zongyuan. Research on several algorithms of person re-identification based on metric learning[D]. Changzhou: Changzhou University, 2018.
[22] GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]. IEEE Conference on Computer Vision and Pattern Recognition, Columbus, 2013: 580-587.
[23] 王金. 基于度量学习的行人重识别方法研究[D]. 武汉: 华中科技大学, 2017. WANG Jin. Research on person re-identification method based on metric learning[D]. Wuhan: Huazhong University of Science and Technology, 2017.
[24] TAO Dapeng,JIN Lianwen,WANG Yongfei,et al. Person reidentification by minimum classification error-based KISS metric learning[J]. IEEE Transactionson Cybernetics,2015,45(2):242-252. DOI: 10.1109/TCYB.2014.2323992
-
期刊类型引用(7)
1. 潘宏春. 基于AR技术的钻探开采事故处理过程安全监控方法研究. 微型电脑应用. 2020(05): 156-159 . 
百度学术
2. 曹永刚. 煤矿安全隐患采集决策智能系统的设计研究. 自动化应用. 2020(04): 119-121 . 百度学术
3. 张强，吴付祥，惠立锋. 煤矿防尘设备在线监控系统的研究. 矿山机械. 2018(09): 58-62 . 百度学术
4. 刘建，郜彤，刘传安. 煤矿安监执法信息管理平台设计研究. 煤矿安全. 2018(03): 245-248 . 百度学术
5. 郜彤，刘传安. 基于GIS的矿井综合信息导航系统. 煤矿安全. 2018(04): 245-248 . 百度学术
6. 陆铮. 矿井瓦斯动态巡检与管控系统设计与应用. 煤炭科学技术. 2018(08): 125-129 . 百度学术
7. 陈梓华，李敬兆. 煤矿安全隐患智能语义采集与智慧决策支持系统. 工矿自动化. 2018(11): 7-13 . 本站查看
其他类型引用(2)
计量
- 文章访问数: 1480
- HTML全文浏览量: 91
- PDF下载量: 50
- 被引次数: 9
