作者：王玉萍, 曾毅

作者单位：郑州科技学院信息工程学院，河南 郑州 450064

关键词：红外图像;行人检测;人类视觉机制;高斯拉普拉斯滤波器;OCS-LBP特征;随机蕨

摘要：

为提高行人检测系统在红外场景中的检测率以及速度，提出一种基于人类视觉机制与ROI融合的红外行人检测方法。根据人类视觉机制来改进LoG滤波抑制背景噪声，通过对滤波后的图像应用ROI融合得到行人候选区域，使提取到的ROI更为准确。另外，提出一种改进的纹理特征OCS-LBP(oriented center symmetric local binary patterns)，对得到的行人候选区域提取HOG特征和OCS-LBP特征，使用随机蕨分类器来进行分类，提升检测的速度与精度。该方法通过与流行的检测算法比较，检测准确率与召回率分别提升7.9%与10.3%，且实时性有较大的提升，具有一定的研究和实用价值。

Pedestrian detection in infrared images using ROI fusion and human visual mechanism
WANG Yuping, ZENG Yi
School of Information Engineering, Zhengzhou University of Science & Technology, Henan 450064, China
Abstract: In order to improve the robustness and the speed of pedestrian detection system in infrared scene, a pedestrian detection method in infrared image based on human visual mechanism and ROI fusion is proposed. According to the human visual mechanism, improved LoG filtering is used to suppress background noise. And a pedestrian candidate region is obtained by applying ROI fusion to the filtered image. What’s more, an improved low-dimension texture feature OCS-LBP is proposed. The HOG feature and OCS-LBP feature are extracted from the obtained pedestrian candidate region, and finally the random fern classifier is used for classification. Compared with the popular detection algorithm, detection accuracy and recall rate of the proposed method are improved by 7.9% and 10.3%, respectively, and the real-time performance is greatly improved. It has certain research and practical value.
Keywords: infrared image;pedestrian detection;human visual mechanism;LoG filter;OCS-LBP feature;random ferns
2021, 47(9):87-93  收稿日期: 2020-10-27;收到修改稿日期: 2020-11-04
基金项目: 河南省教育科学"十三五"规划2020年度一般课题( 2020YB0335)
作者简介: 王玉萍(1979-),女,河南焦作市人,副教授,硕士,主要从事机器视觉与虚拟现实方面的研究
参考文献
[1] 刘桂雄, 张瑜, 蔡柳依婷. 机器视觉检测图像拼接融合技术研究进展[J]. 中国测试, 2020, 46(1): 1-6
[2] YOO J, HWANG S, KIM S D, et al. Scale-invariant template matching using histogram of dominant gradients[J]. Pattern Recognit, 2014, 47(9): 3006-3018
[3] ZHAO J, FENG H, XU Z, et al. Real-time automatic small target detection using saliency extraction and morphological theory[J]. Optics & Laser Technology, 2013, 47(4): 268-277
[4] 孟晓燕, 段建民, 刘丹. 采用人体树图与混合粒子群聚类的行人检测[J]. 光学精密工程, 2018, 26(7): 1802-1812
[5] LI J, GONG W, LI W, et al. Robust pedestrian detection in thermal infrared imagery using the wavelet transform[J]. Infrared Physics and Technology, 2010, 53(4): 267-273
[6] LI W, ZHENG D, Zhao T, et al. An effective approach to pedestrian detection in thermal imagery[C]//Eighth International Conference on Natural Computation. IEEE, 2012.
[7] 宋子浩, 吴斌, 周挺. 输入受限情况下的球形机器人轨迹跟踪控制[J]. 中国测试, 2020, 46(5): 120-126
[8] KIM S, YANG Y, LEE J, et al. Small target detection utilizing robust methods of the human visual system for IRST[J]. Journal of Infrared Millimeter & Terahertz Waves, 2009, 30(9): 994-1011
[9] WANG Y, ZHAO Q, WANG B, et al. A real-time active pedestrian tracking system inspired by the human visual system[J]. Cognitive Computation, 2016, 8(1): 39-51
[10] DENG H, SUN X, LIU M, et al. Entropy-based window selection for detecting dim and small infrared targets[J]. Pattern Recognition, 2017, 61(2): 66-77
[11] LI Y, ZUO M J, LIN J, et al. Fault detection method for railway wheel flat using an adaptive multiscale morphological filter[J]. Mechanical Systems and Signal Processing, 2017, 84: 642-658
[12] ANTONIO F, MARIA L, JUAN S. Thermal-Infrared Pedestrian ROI Extraction through Thermal and Motion Information Fusion[J]. Sensors, 2014, 14(4): 6666-6676
[13] KWAK J Y, KO B C, NAM J Y. Pedestrian tracking using online boosted random ferns learning in far-infrared imagery for safe driving at night[J]. IEEE Transactions on Intelligent Transportation Systems, 2016, 18(1): 69-81
[14] 曹鹏霞, 李文新, 马伟苹. 基于改进随机蕨的跟踪注册方法研究[J]. 计算机应用与软件, 2019, 36(11): 51-56