多特征多阈值级联AdaBoost行人检测器

崔华; 张骁; 郭璐; 袁超; 薛世焦; 宋焕生

doi:10.19818/j.cnki.1671-1637.2015.02.012

多特征多阈值级联AdaBoost行人检测器

doi: 10.19818/j.cnki.1671-1637.2015.02.012

长安大学信息工程学院, 陕西西安 710064

基金项目:

国家863计划项目 2012AA112312

详细信息

作者简介:
崔华(1977-), 女, 河南南阳人, 长安大学副教授, 工学博士, 从事交通图像识别研究

中图分类号: U491.6
计量
- 文章访问数: 739
- HTML全文浏览量: 153
- PDF下载量: 901
- 被引次数: 0
出版历程
- 收稿日期: 2014-11-13
- 刊出日期: 2015-02-25

Cascade AdaBoost pedestrian detector with multi-features and multi-thresholds

School of Information Engineering, Chang'an University, Xi'an 710064, Shaanxi, China

More Information

Author Bio:
CUI Hua(1977-), female, associate professor, PhD, +86-29-62630027, huacui@chd.edu.cn

摘要

摘要: 为了满足更快、更准、更鲁棒的行人检测需求, 考虑交通监控视频图像质量不高与局部特征不明显的缺点, 采用简单的行人特征来实现行人检测。除矩形度、高宽比、轮廓复杂度、宽度比、行人面积特征外, 特定选用了对遮挡等干扰具有强鲁棒性的头部圆形度这一简单的局部特征。考虑交通监控视频图像中行人的尺寸变化, 引入区域划分策略划分图像区域。兼顾高检测率和低误检率, 根据分类误差最小原则与正样本分类率最大原则训练多个单特征多阈值AdaBoost行人检测器。为了优化多个行人检测器级联后的检测性能, 在兼顾检测性能和检测速度的基础上, 定义了以贡献率作为行人检测器的级联规则, 依据贡献率大小确定的级联次序为基于高宽比、宽度比、矩形度、行人面积、轮廓复杂度和头部圆形度的行人检测器, 依次进行级联, 建立了新的多特征多阈值级联AdaBoost行人检测器。选用3个交通场景对行人检测器进行测试, 并与单级AdaBoost行人检测器与现有2种级联AdaBoost行人检测器进行比较。分析结果表明: 在3个交通场景的检测中, 相比其他几种行人检测器, 多特征多阈值级联AdaBoost行人检测器具有较高检测率、较快的检测速度和较低误检率, 检测率最低为96.70%, 误检率最高为0.67%, 检测时间小于5s, 满足交通场景中对行人检测实时性和可靠性的要求。
- 交通图像处理 /
- 行人检测 /
- 特征提取 /
- AdaBoost分类器 /
- 区域划分 /
- 级联规则
Abstract: In order to meet the practical demand for pedestrian detection with high speed, high accuracy and strong robustness, in view of the poor quality and unapparent local image features of traffic videos, some simple pedestrian features were chosen for pedestrian detection.Besides rectangle degree, ratio of height to width, shape complexity, normalized width, and pedestrian area, head density was applied because it is a simple local feature and has strong robustness for occlusion interference.Considering the size changing of pedestrian in the image, region division strategy was introduced into image region division.An improved training algorithm based on the minimum principle of classification error and the maximum principle of positive sample classification rate was implemented by considering both high detection rate and low false detection rate, thus several single-feature AdaBoost pedestrian detectors with multi-thresholds were obtained.To optimize the detection performance of cascade pedestrian detectors, the cascade rulewas obtained in term of the contribution rate.The contribution rate was defined by weighing detection performance and detection speed.The cascade order was the detectors based on ratio of height to width, normalized width, rectangle degree, pedestrian area, shape complexity and head density.The pedestrian detectors were sequentially cascaded according to the cascade order, thus a cascade AdaBoost pedestrian detector with multi-features and multi-thresholds was constructed.The proposed pedestrian detector was tested by using 3traffic scenes, and compared with singlecascade-level AdaBoost pedestrian detector and 2existed cascade AdaBoost pedestrian detectors.Analysis result indicates that in 3traffic scenes, compared with the other pedestrain detectors, the proposed pedestrain detector has higher detection rate, higher detection speed and lower false detection rate, the minimum detection rate is 96.70%, the maximum false detection rate is0.67%, and the detection time is less than 5s.So the detector satisfies the real-time and reliable requirements of pedestrian detection in traffic scene.
- traffic image processing /
- pedestrian detection /
- feature extraction /
- AdaBoost classifier /
- region division /
- cascade rule

HTML全文

图 1 行人头部区域提取过程

Figure 1. Extraction process of pedestrian head region

下载: 全尺寸图片幻灯片

图 2 图像区域划分

Figure 2. Image region separation

下载: 全尺寸图片幻灯片

图 3 级联检测器的结构

Figure 3. Structure of cascade detector

下载: 全尺寸图片幻灯片

图 4 多交通场景下的行人检测结果

Figure 4. Pedestrian detection results in various traffic scenes

下载: 全尺寸图片幻灯片

图 5 行人检测器检测结果

Figure 5. Detection results of pedestrian detectors

下载: 全尺寸图片幻灯片

表 1 行人检测器测试结果比较

Table 1. Test result comparison of pedestrian detectors

参考文献(24)

[1]	郭立君, 刘曦, 赵杰煜, 等. 结合运动信息与表观特征的行人检测方法[J]. 软件学报, 2012, 23(2): 299-309. https://www.cnki.com.cn/Article/CJFDTOTAL-RJXB201202012.htm GUO Li-jun, LIU Xi, ZHAO Jie-yu, et al. Pedestrian detection method of integrated motion information and appearance features[J]. Journal of Software, 2012, 23(2): 299-309. (in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-RJXB201202012.htm
[2]	NEGRI P, GOUSSIES N, LOTITO P. Detecting pedestrians on a movement feature space[J]. Pattern Recognition, 2014, 47(1): 56-71. doi: 10.1016/j.patcog.2013.05.020
[3]	DOLLÁR P, WOJEK C, SCHIELE B, et al. Pedestrian detection: an evaluation of the state of the art[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2012, 34(4): 743-761. doi: 10.1109/TPAMI.2011.155
[4]	BORGES P V K. Pedestrian detection based on blob motion statistics[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2013, 23(2): 224-235. doi: 10.1109/TCSVT.2012.2203217
[5]	GAVRILA D M. A Bayesian, exemplar-based approach to hierarchical shape matching[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2007, 29(8): 1408-1421. doi: 10.1109/TPAMI.2007.1062
[6]	PRIOLETTI A, MØGELMOSE A, GRISLERI P, et al. Part-based pedestrian detection and feature-based tracking for driver assistance: real-time, robust algorithms, and evaluation[J]. IEEE Transactions on Intelligent Transportation Systems, 2013, 14(3): 1346-1359. doi: 10.1109/TITS.2013.2262045
[7]	ENZWEILER M, GAVRILA D M. Monocular pedestrian detection: survey and experiments[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2009, 31(12): 2179-2195.
[8]	VÁZQUEZ D, LÓPEZ A M, MARÍN J, et al. Virtual and real world adaptation for pedestrian detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2014, 36(4): 797-809.
[9]	CAO Xian-bin, WANG Zhong, YAN Ping-kun, et al. Transfer learning for pedestrian detection[J]. Neurocomputing, 2013, 100: 51-57.
[10]	CHENG Wen-chang, JHAN D M. A self-constructing cascade classifier with AdaBoost and SVM for pedestrian detection[J]. Engineering Applications of Artificial Intelligence, 2013, 26(3): 1016-1028.
[11]	汤义. 智能交通系统中基于视频的行人检测与跟踪方法的研究[D]. 广州: 华南理工大学, 2010. TANG Yi. Study on video-based detection and tracking method of pedestrian in ITS[D]. Guangzhou: South China University of Technology, 2010. (in Chinese).
[12]	DALAL N, TRIGGS B. Histograms of oriented gradients for human detection[C]∥IEEE. Proceedings of the 2005IEEE Computer Society Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2005: 886-893.
[13]	邵枭虎. 基于AdaBoost与显著信息的行人检测算法[D]. 成都: 电子科技大学, 2012. SHAO Xiao-hu. The algorithm of pedestrian detection based on AdaBoost and saliency information[D]. Chengdu: University of Electronic Science and Technology of China, 2012. (in Chinese).
[14]	MEYNET J, POPOVICI V, THIRAN J P. Face detection with boosted Gaussian features[J]. Pattern Recognition, 2007, 40(8): 2283-2291.
[15]	TIAN Hong, DUAN Zhu, ABRAHAM A, et al. A novel multiplex cascade classifier for pedestrian detection[J]. Pattern Recognition Letters, 2013, 34(14): 1687-1693.
[16]	常峰, 杨彬, 窦建华. 基于多特征和级联分类器的行人检测算法[J]. 合肥工业大学学报: 自然科学版, 2014, 37(12): 1456-1461. https://www.cnki.com.cn/Article/CJFDTOTAL-HEFE201412010.htm CHANG Feng, YANG Bin, DOU Jian-hua. Pedestrian detection algorithm with multiple feature and cascade classifier[J]. Journal of Hefei University of Technology: Natural Science, 2014, 37(12): 1456-1461. (in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-HEFE201412010.htm
[17]	李梦涵, 熊淑华, 熊文, 等. 多尺度级联行人检测算法的研究与实现[J]. 计算机技术与发展, 2014, 24(8): 10-13. https://www.cnki.com.cn/Article/CJFDTOTAL-WJFZ201408003.htm LI Meng-han, XIONG Shu-hua, XIONG Wen, et al. Research and realization of pedestrian detection algorithm by multi-scale cascaded features[J]. Computer Technology and Development, 2014, 24(8): 10-13. (in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-WJFZ201408003.htm
[18]	程如中, 赵勇, 王执中, 等. 实时行人检测预警系统[J]. 交通运输工程学报, 2012, 12(5): 110-118, 126. http://transport.chd.edu.cn/article/id/201205015 CHENG Ru-zhong, ZHAO Yong, WONG Chup-chung, et al. Real-time pedestrian detecting and warning system[J]. Journal of Traffic and Transportation Engineering, 2012, 12(5): 110-118, 126. (in Chinese). http://transport.chd.edu.cn/article/id/201205015
[19]	FREUND Y, SCHAPIRE R E. A decision-theoretic generalization of on-line learning and an application to Boosting[J]. Journal of Computer and System Sciences, 1997, 55(1): 119-139.
[20]	SCHAPIRE R E, SINGER Y. Improved boosting algorithms using confidence-rated predictions[J]. Machine Learning, 1999, 37(3): 297-336.
[21]	GERÓNIMO D, LÓPEZ A M, SAPPA A D, et al. Survey of pedestrian detection for advanced driver assistance systems[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2010, 32(7): 1239-1258.
[22]	DENG Jin-hao, ZHU Juan. Research on pedestrian detection algorithms based on video[C]∥IEEE. 2010IEEE Conference on Computer Design and Applications. New York: IEEE, 2010: 474-478.
[23]	LEE Y D, LI Z Z, ZHANG S R, et al. Safety impacts of red light running photo enforcement at urban signalized intersections[J]. Journal of Traffic and Transportation Engineering: English Edition, 2014, 1(5): 309-324.
[24]	LIU Qing-hua, CHUNG E, ZHAI Liu-jia. Fusing moving average model and stationary wavelet decomposition for automatic incident detection: case study of Tokyo expressway[J]. Journal of Traffic and Transportation Engineering: English Edition, 2014, 1(6): 404-414.