连续跟踪状态下基于可分性特征的目标优化分类

doi:10.11772/j.issn.1001-9081.2014.05.1275

计算机应用 ›› 2014, Vol. 34 ›› Issue (5): 1275-1278.DOI: 10.11772/j.issn.1001-9081.2014.05.1275

连续跟踪状态下基于可分性特征的目标优化分类

李志华¹,²,李秋峦²

1. 浙江大学数字技术及仪器研究所，杭州 310027
2. 杭州师范大学国际服务工程学院，杭州 311100

收稿日期:2013-11-15 修回日期:2013-12-30 发布日期:2014-05-30 出版日期:2014-05-01
通讯作者: 李志华
作者简介:李志华(1981-)，男，江西南昌人，讲师，博士，主要研究方向：图像识别、嵌入式系统；李秋峦(1990-)，男，浙江金华人，硕士研究生，主要研究方向：智能识别、计算机网络。
基金资助:
国家自然科学基金资助项目

Object classification based on discriminable features and continuous tracking

LI Zhihua¹,²,LIU Qiuluan²

1. Institute of Advanced Digital Technology and Instrument, Zhejiang University, Hangzhou Zhejiang 310027, China
2. Institute of Service Engineering, Hangzhou Normal University, Hangzhou Zhejiang 311100, China;

Received:2013-11-15 Revised:2013-12-30 Online:2014-05-30 Published:2014-05-01
Contact: LI Zhihua

摘要/Abstract

摘要：

针对拥塞复杂监控场景中目标的准确分类问题，提出了一种连续跟踪状态下基于可分性特征的目标优化分类方法。首先对整个场景中所有目标提取简单的颜色、形状和位置特征建立初始目标匹配，利用目标的运动方向及速率预测下帧中优先搜索区域以提高目标匹配效率，减少运算量，并对未建立对应关系的遮挡目标采用外观特征模型进行再匹配。为了提高目标分类的准确率，系统利用连续跟踪状态下目标特征的不间断提取和匹配，根据匹配最大概率决定最优分类结果。通过多种场景的实验结果表明，该方法的分类准确度比未利用连续跟踪信息的方案获得了更好分类准确度，平均达到了97%，有效改善了复杂场景中目标分类精度。

Abstract:

Aiming at object classification problem in heavily crowded and complex visual surveillance scenes, a real-time object classification approach was proposed based on discriminable features and continuous tracking. Firstly rapid features matching including color, shape and position was utilized to build the initial target correspondence in the whole scene, in which motion direction and velocity of the moving target were used to predict the preferable searching area in the next frame to accelerate the target matching process. And then the appearance model was utilized to rematch the occluded object without establishing the correspondence. In order to enhance the classification precision, the final object classification results were determined by the maximum probability of continuous object feature extraction and classification according to the tracking results. Experimental results show that the proposed method gets better classification precision compared with the method which do not utilized the continuous tracking，and its correct rate averagely reaches 97%. The new scheme effectively improves the performance of object classification in the complex scenes.

中图分类号:

TP391.4

李志华李秋峦. 连续跟踪状态下基于可分性特征的目标优化分类[J]. 计算机应用, 2014, 34(5): 1275-1278.

LI Zhihua LIU Qiuluan. Object classification based on discriminable features and continuous tracking[J]. Journal of Computer Applications, 2014, 34(5): 1275-1278.

参考文献

［1］LIU J, WANG Y, ZHANG Z, et al. Multi-view moving object classification via transfer learning ［C］// Proceedings of the 1st Asian Conference on Pattern Recognition. Piscataway: IEEE Press, 2011:259-263.
［2］LUO Y, TAO D, GENG B, et al. Manifold regularized multitask learning for semi-supervised multilabel image classification ［J］. IEEE Transactions on Image Processing, 2013,22(2):523-536.
［3］GOMEZ-CHOVA L, CAMPS-VALLS G, BRUZZONE L, et al. Mean map kernel methods for semi-supervised cloud classification ［J］. IEEE Transactions on Geoscience and Remote Sensing, 2010,48(1):207-220.
［4］GONG Y C, LIU F, CHEN C. Graph-based semi-supervised learning with manifold preprocessing for image classification ［C］// Proceedings of the 2008 IEEE International Conference on Systems, Man and Cybernetics. Piscataway: IEEE Press, 2008:391-395.
［5］BAI Y, WANG X. Graph-based semi-supervised method for image classification in combination with mean shift ［J］. Journal of Computer Applications, 2013,33(9):2606-2609.(白艺娜，汪西莉.结合均值漂移的基于图的半监督图像分类［J］.计算机应用，2013,33(9):2606-2609.)
［6］LIU W, HE J, CHANG S F. Large graph construction for scalable semi-supervised learning ［EB/OL］. ［2012-10-10］. http://machinelearning.wustl.edu/mlpapers/paper_files/icml2010_LiuHC10.pdf.
［7］HU W, TAN T, WANG L, et al. A survey on visual surveillance of object motion and behaviors ［J］. IEEE Transactions on Systems, Man, and Cybernetics, Part C: Applications and Reviews, 2004,34(3):334-352.
［8］JIANG Y, PENG Q. The mining of stable image feature-compositions in object recognition ［J］. Journal of Image and Graphics, 2012,17(1):99-105.(姜永兵,彭启民.目标识别中的稳定图像特征组合发掘［J］.中国图象图形学报,2012,17(1):99-105.)
［9］LI Z, XIAO G, DAI Y, et al. Moving target classification using SVM probability and post-filtering ［J］. Application Research of Computers, 2010,27(2):778-790.(李占闯,肖国强,代毅,等.基于SVM输出概率和后置滤波的运动目标分类［J］.计算机应用研究, 2010,27(2):778-790.)
［10］VIOLA P, JONES M J, SNOW D. Detecting pedestrians using patterns of motion and appearance ［J］. International Journal of Computer Vision, 2005,63(2):153-161.
［11］JAVED O,SHAH M. Tracking and object classification for automated surveillance ［C］// Proceedings of the 7th European Conference on Computer Vision. Berlin: Springer-Verlag, 2002:343-357.
［12］BROWN L M. View independent vehicle/person classification ［C］// Proceedings of the ACM 2nd International Workshop on Video Surveillance and Sensor Networks. New York: ACM, 2004:114-123.
［13］STAUFFER C, TIEU K, LEE L. Robust automated planar normalization of tracking data ［EB/OL］. ［2013-01-23］. http://www.researchgate.net/publication/2908889_Robust_Automated_Planar_Normalization_of_Tracking_Data/file/50463518e9b3e5160b.pdf.
［14］TSUCHIYA M, FUJIYOSHI H. Evaluating feature importance for object classification in visual surveillance ［C］// Proceedings of the 18th IEEE International Conference on Pattern Recognition. Piscataway: IEEE Press, 2006:978-981.

[1]	张佳慧李晓明张嘉祥. 强化形态感知的路面缺陷检测算法[J]. 《计算机应用》唯一官方网站, 0, (): 0-0.
[2]	杨建锋陈斌李雨轩. 基于点云重构的自监督点云异常检测方法[J]. 《计算机应用》唯一官方网站, 0, (): 0-0.
[3]	蒋畅江向杰何旭颖. 面向机械臂抓取的双目视觉目标定位算法[J]. 《计算机应用》唯一官方网站, 0, (): 0-0.
[4]	边小勇胡其仁袁培洋. 多注意力对比学习的红外小目标检测[J]. 《计算机应用》唯一官方网站, 0, (): 0-0.
[5]	李钟华钟庚辛范萍朱恒亮. 通过边界挖掘和背景引导的伪装目标检测[J]. 《计算机应用》唯一官方网站, 0, (): 0-0.
[6]	吴松霖张广朝姚远彭博. 基于判别区域引导的多视图困难气道识别[J]. 《计算机应用》唯一官方网站, 0, (): 0-0.
[7]	李强白少雄熊源袁薇. 基于视觉大模型隐私保护的监控图像定位[J]. 《计算机应用》唯一官方网站, 0, (): 0-0.
[8]	况世雄姚俊波陆佳炜王琪冰肖刚. 基于动态图卷积网络的电梯乘客异常行为数据增强方法[J]. 《计算机应用》唯一官方网站, 0, (): 0-0.
[9]	康斌陈斌王俊杰李昱林赵军智咸伟志. 基于多粒度共享语义中心关联的文本到人物检索方法[J]. 《计算机应用》唯一官方网站, 0, (): 0-0.
[10]	王磊胡节彭博. 用于半监督火灾检测的分布自适应和动态课程伪标签框架[J]. 《计算机应用》唯一官方网站, 0, (): 0-0.
[11]	王子怡李卫军刘雪洋丁建平刘世侠苏易礌. 基于Swin Transformer与多尺度特征融合的图像描述方法#br# [J]. 《计算机应用》唯一官方网站, 0, (): 0-0.
[12]	熊炜, 陈奕博, 张丽真, 杨茜, 邹勤. 利用多帧序列影像的自监督单目深度估计[J]. 《计算机应用》唯一官方网站, 2024, 44(12): 3907-3914.
[13]	付可意, 王高才, 邬满. 基于改进区域提议网络和特征聚合小样本目标检测方法[J]. 《计算机应用》唯一官方网站, 2024, 44(12): 3790-3797.
[14]	杨成昊, 胡节, 王红军, 彭博. 基于注意力机制的不完备多视图聚类算法[J]. 《计算机应用》唯一官方网站, 2024, 44(12): 3784-3789.
[15]	方鹏, 赵凡, 王保全, 王轶, 蒋同海. 区块链3.0的发展、技术与应用[J]. 《计算机应用》唯一官方网站, 2024, 44(12): 3647-3657.

连续跟踪状态下基于可分性特征的目标优化分类

Object classification based on discriminable features and continuous tracking

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics