首页 > 论文 > 应用光学 > 39卷 > 6期(pp:832-838)

改进Fast-RCNN的双目视觉车辆检测方法

Binocular vision vehicle detection method based on improved Fast-RCNN

  • 摘要
  • 论文信息
  • 参考文献
  • 被引情况
  • PDF全文
分享:

摘要

针对不同空间尺度的车辆表现出显著不同的特征导致检测算法效率低、准确性差且单目难以准确获取车辆距离信息的问题, 提出了一种改进Fast-RCNN的汽车目标检测法, 利用双目视觉对车辆进行测距。首先利用双目立体相机采集前方图像并进行预处理, 加载深度神经网络Fast-RCNN的训练数据, 再针对汽车不同空间尺度引入多个内置的子网络, 将来自所有子网络的输出自适应组合对车辆进行检测, 然后利用SURF特征匹配算法进行左右图像的立体匹配, 根据匹配数据进行三维重建并确定车辆质心坐标, 从而测量出车辆与双目相机之间的距离。实验结果表明, 所述算法可以实现对车辆的快速检测, 检测时间比传统的Fast-RCNN缩短了42 ms, 并且实现了对5 m范围车辆距离的准确测量, 其误差仅为2.4%, 精确度高, 实时性好。

Abstract

Vehicles with different spatial scales exhibit significantly different characteristics, resulting in low efficiency, poor accuracy of vehicle detection methods and difficulty for accurately obtaining vehicle distance information. To solve this problem, a vehicle detection method based on improved fast region-based convolutional neural network (Fast-RCNN) to detect vehicle targets was proposed,which using binocular vision to range vehicles. Firstly the binocular stereo camera was used to acquire the image information and perform preprocessing, the training data of the deep neural network Fast-RCNN was loaded, then multiple built-in sub networks were introduced for different spatial scales of vehicles, and the output of all sub networks was adaptively combined to detect vehicles. At last the speeded up robust features (SURF) matching algorithm was utilized to carry out the stereo matching of the left and right images.And based on the matching data, the 3D reconstruction and vehicle centroid coordinates were determined so as to measure the distance between vehicle and binocular camera. Experimental results show that the algorithm can achieve fast detection of vehicles, the detection time is shorter than the traditional Fast-RCNN by 42 ms and the accurate measurement of vehicle distance in 5 m can be achieved with an error of only 2.4%. The proposed method has high accuracy and good real-time performance.

Newport宣传-MKS新实验室计划
补充资料

中图分类号:TN206;TP391.4

DOI:10.5768/jao201839.0602001

所属栏目:光电信息获取与处理

基金项目:国家自然科学基金(51475387); 四川省科技支撑项目(2016GZ0026, 2017HH0102, 2017GZ0164, 2018GZ0388)

收稿日期:2018-06-21

修改稿日期:2018-08-07

网络出版日期:--

作者单位    点击查看

张 琦:西南交通大学 机械工程学院,四川 成都 610031
胡广地:西南交通大学 机械工程学院,四川 成都 610031
李雨生:西南交通大学 机械工程学院,四川 成都 610031
赵鑫:西南交通大学 机械工程学院,四川 成都 610031

联系人作者:张琦(15708455266@163.com)

备注:张琦(1994-),男,山西运城人,硕士,主要从事机器视觉和自动驾驶方面的研究工作。

【1】JALALAT M, NEJATI M, MAJIDI A. Vehicle detection and speed estimation using cascade classifier and sub-pixel stereo matching[C]// Signal Processing and Intelligent Systems.USA: IEEE, 2017: 1-5.

【2】WEN X, SHAO L, FANG W, et al. Efficient Feature selection and classification for vehicle detection[J]. IEEE Transactions on Circuits & Systems for Video Technology, 2015, 25(3): 508-517.

【3】XU Guoyan, WANG Chuanrong, GAO Feng, et al. Area determination algorithm for vehicle video detection[J]. Journal of Beijing University of Aeronautics and Astronautics, 2010, 36(7): 781-784.
徐国艳, 王传荣, 高峰,等. 车辆视频检测感兴趣区域确定算法[J]. 北京航空航天大学学报, 2010, 36(7): 781-784.

【4】ROSEBROCK D, RILK M. Real-time vehicle detection with a single camera using shadow segmentation and temporal verification[C]// IEEE/RSJ International Conference on Intelligent Robots and Systems.USA: IEEE, 2012: 3061-3066.

【5】LEE J, KIM T, LEE S, et al. Vehicle tracking using templete matching based on feature point[C].USA: IEEE,2005.

【6】TALUKDER A, MATTHIES L. Real-time detection of moving objects from moving vehicles using dense stereo and optical flow[C]// IEEE/RSJ International Conference on Intelligent Robots and Systems.USA: IEEE, 2004: 3718-3725.

【7】CHEN Z, YING C, LIN C, et al. Multi-view vehicle type recognition with feedback-enhancement multi-branch CNNs[C].USA: IEEE, 2017.

【8】LUO W, SCHWING A G, URTASUN R. Efficient deep learning for stereo matching[C]// IEEE Conference on Computer Vision and Pattern Recognition.USA: IEEE, 2016: 5695-5703.

【9】GU Zhaolun. Vehicle detection and distance measurement based on smart phone[D]. Hangzhou : Zhejiang University, 2017.
顾兆伦. 基于智能手机的车辆检测与车距测量[D]. 杭州: 浙江大学, 2017.

【10】ARBELáEZ P, PONT-TUSET J, BARRON J, et al. Multiscale combinatorial grouping[C]// Computer Vision and Pattern Recognition. USA: IEEE, 2014: 328-335.

【11】TONG Zhixue, ZHAO Tao, WANG Xiaowei. Range location and vehicle speed estimation based on binocular sequence images[J]. Journal of Applied Optics, 2017, 38(5): 764-769.
同志学, 赵涛, 王消为. 基于双目序列图像的测距定位与自车速度估计[J]. 应用光学, 2017, 38(5): 764-769.

【12】BAY H, ESS A, TUYTELAARS T, et al. Speeded-up robust features (SURF)[J]. Computer Vision & Image Understanding, 2008, 110(3): 346-359.

【13】ZHANG B, JIAO Y, MA Z, et al. An efficient image matching method using speed Up robust features[C]// IEEE International Conference on Mechatronics and Automation.USA: IEEE, 2014: 553-558.

引用该论文

Zhang Qi,Hu Guangdi,Li Yusheng,Zhao Xin. Binocular vision vehicle detection method based on improved Fast-RCNN[J]. Journal of Applied Optics, 2018, 39(6): 832-838

张 琦,胡广地,李雨生,赵鑫. 改进Fast-RCNN的双目视觉车辆检测方法[J]. 应用光学, 2018, 39(6): 832-838

被引情况

【1】陈苏婷,郭子烨,张艳艳. 基于局部哈希学习的大面阵CCD航拍图像匹配方法. 应用光学, 2019, 40(2): 259-264

您的浏览器不支持PDF插件,请使用最新的(Chrome/Fire Fox等)浏览器.或者您还可以点击此处下载该论文PDF