深圳大学学报理工版

刘翠响,袁香伟,王宝珠,等.最小均衡化后的行人重识别[J].深圳大学学报理工版,2019,36(No.4(347-472)):447-452.[doi:10.3724/SP.J.1249.2019.04447]
LIU Cuixiang,YUAN Xiangwei,WANG Baozhu,et al.Minimum equalization for pedestrain re-identification[J].Journal of Shenzhen University Science and Engineering,2019,36(No.4(347-472)):447-452.[doi:10.3724/SP.J.1249.2019.04447]

点击复制

最小均衡化后的行人重识别

刘翠响,袁香伟,王宝珠,张亚凤,马杰

河北工业大学电子信息工程学院,天津 300401

关键词：人工智能; 行人重识别; 特征融合; 下采样; 交叉二次判别分析度量学习; 直方图均衡化; 图像处理; 模式识别

Minimum equalization for pedestrain re-identification

LIU Cuixiang, YUAN Xiangwei, WANG Baozhu, ZHANG Yafeng, and MA Jie

School of Electronic and Information Engineering, Hebei University of Technology, Tianjin 300401, P. R. China

Keywords： artificial intelligence; pedestrain re-identification; feature fusion; down sampling; cross-view quadratic discriminant analysis(XQDA)measures learning; histogram equalization; image processing; pattern recognition

DOI: 10.3724/SP.J.1249.2019.04447

备注

摘要

全文

图/表

参考文献

为解决实际监控场景中的行人重识别技术的智能应用,考虑到行人图像拍摄角度不断变化的情况,将颜色和纹理等特征进行融合,利用部分局部块提取图像特征; 针对行人轮廓不清晰,提出在纹理特征提取前实现直方图均衡化的方法; 通过对图像进行两次下采样,使算法具有更好的比例尺度不变性.与现有的局部最大概率(local maximal occurrence, LOMO)特征与跨视图二次鉴别分析(cross-view quadratic discriminant analysis, XQDA)方法结合的重识别方法进行对比,结果表明,在数据集VIPeR、PKU-Reid和i-LIDS-VID上重识别率rank1分别提高了0.28%、1.75%和0.20%,证明采用最小均衡化后的行人重识别率得到了提升.

In order to solve the intelligent application of pedestrain re-identification technology in the actual monitoring scenes, the color and texture features are fused and some local blocks are used to extract image features by considering the changing shooting angle of pedestrian image. A method of histogram equalization before texture feature extraction is proposed to solve the problem of unclear pedestrian contour. By down sampling the image twice, the algorithm has better scale invariance. Compared with the existing re-identification method of cross-view quadratic discriminant analysis(XQDA)combined with local maximal occurrence(LOMO)characteristics, the experimental results show that the corresponding re-identification rate rank1 on the datasets of VIPeR, PUK-Reid and i-LIDS-VID is improved by 0.28%, 1.75% and 0.20%, respectively, which proves that the recognition rate of pedestrain re-identification with minimum equalization is improved.

引言
1 图像均衡化基本理论
2 行人重识别算法
3 实验结果与分析
4 结语

[1] 张耿宁,王家宝,李阳,等.基于特征融合与核局部Fisher判别分析的行人重识别[J].计算机应用,2016,36(9):2597-2600, 2635.
[2] 张华.基于空间颜色特征的行人重识别方法[J].华中科技大学学报自然科学版,2013,41(增刊Ⅱ):209-212,217.
[3] 孙金玉,王洪元,张继,等.基于块稀疏表示的行人重识别方法[J].计算机应用,2018,38(2):448- 453.
[4] BAK S, CORVEE E, BRÉMOND F, et al. Person re-identification using Haar-based and DCD-based signature[C]// Proceedings of the 7th IEEE International Conference on Advanced Video and Signal Based Surveillance. Washington D C: IEEE Computer Society Washington, 2010: 1-8.
[5] LIAO Shengcai, HU Yang, ZHU Xiangyu, et al. Person re-identification by local maximal occurrence representation and metric learning[C]// The IEEE International Conference on Computer Vision and Pattern Recognition. Boson, USA: IEEE, 2015: 2197- 2206.
[6] MATSUKAWA T, OKABE T, SUZUKI E, et al. Hierarchical Gaussian descriptor for person re-identification[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, USA: IEEE, 2016: 1363- 1372.
[7] XING E P, NG A Y, JORDAN M I, et al. Distance metric learning, with application to clustering with side-information[C]// Proceedings of the 15th International Conference on Neural Information Processing Systems. Cambridge, USA: MIT Press, 2002: 521-528.
[8] DIKMEN M, AKBAS E, HUANG T S, et al. Pedestrian recognition with a learned metric[C]// Proceedings of the 10th Asian Conference on Computer Vision. Berlin: Springer-Verlag, 2011: 501-512.
[9] ZHENG Weishi, GONG Shaogang, XIANG Tao. Reidentification by relative distance comparison[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013, 35(3): 653- 668.
[10] KÖSTINGER M, HIRZER M, WOHLHART P, et al. Large scale metric learning from equivalence constrains[C]// Proceedings of the 2012 IEEE Conference on Computer Vision and pattern Recognition. Washington D C: IEEE Computer Society, 2012: 2288-2295.
[11] PEDAGADI S, ORWELL J, VELASTINS, et al. Local fisher discriminant analysis for pedestrian re-identification[C]// Proceedings of the 2013 IEEE Conference on Computer Vision and Pattern Recognition. Washington D C: IEEE Computer Society, 2013: 3318-3325.
[12] 刘琦,侯丽.基于辨识特征后融合的行人再识别[J/OL].计算机应用研究,2019,36(9).[2018- 05-24].http://www.arocmag.com/article/02-2019- 09- 054.html
[13] 杨新雨.基于监控场景的行人重识别方法的研究与应用[D].成都:电子科技大学,2018.
[14] 宋丽丽.迁移度量学习行人再识别算法[J/OL].计算机工程与应用.[2018-12-14]. http://kns.cnki.net/kcms/detail/11.2127.TP.20181212.1740.033.html