道路交叉口自动检测算法的研究Research on automatic recognition algorithm of road intersection
王龙飞,刘智,金飞,王番
摘要(Abstract)：
针对高分辨率遥感影像道路交叉口特征不明显、检测难度大等问题,该文提出一种改进的道路交叉口自动检测算法。该算法在YOLOv3网络基础上,首先采用参数修正单元激活卷积层,使目标特征在传递过程中保留更多负信息;然后在特征提取端与特征检测端之间实现多尺度特征融合,增强目标细节特征的提取;最后将单向卷积模块改进为多通道卷积模块,对卷积模块横向拉伸后再纵向聚合。为了验证算法有效性,对常见7种类型交叉口进行测试,实验结果表明:改进后算法对复杂背景下小尺寸道路交叉口的检测效果得到明显提升,有效实现了多种类型的道路交叉口自动化检测。
关键词(KeyWords)： 道路交叉口;YOLOv3网络;参数修正单元;多尺度融合;多通道卷积模块
基金项目(Foundation):
作者(Author): 王龙飞,刘智,金飞,王番
DOI: 10.16251/j.cnki.1009-2307.2020.05.019
参考文献(References)：
- [1]HU J,RAZDAN A,FEMIANI J C,et al.Road network extraction and intersection detection from aerial images by tracking road footprints[J].IEEETransactions on Geoscience and Remote Sensing,2007,45(12):4144-4157．
- [2]蔡红玥，姚国清．高分辨率遥感图像道路交叉口自动提取[J]．国土资源遥感，2016,28(1):63-71.(CAI Hongyue,YAO Guoqing.Automatic extraction of road intersections from high resolution remote sensing images[J].Remote Sensing for Land and Resources,2016,28(1):63-71．）
- [3]闫箐．基于高分遥感影像的道路信息快速提取技术研究[D]．南昌：江西农业大学，2016:1-66.(YAN Jing.Research on rapid extraction of road information based on high score remote sensing image[D].Nanchang:Jiangxi Agricultural University,2016:1-66．）
- [4]李润生，司毅博，德宝，等．基于语义规则的高分辨率影像道路交叉口提取方法[J]．测绘科学技术学报，2017,2:168-174.(LI Runsheng,SI Yibo,DE Bao,et al.Extraction method of high-resolution image road intersection based on semantic rules[J].Journal of Geomatics Science and Technology,2017,2:168-174．）
- [5]DENG L,YU D.Deep learning:methods and applications[J].Foundations&Trends in Signal Processing,2014,7(3):197-387．
- [6]UCAR A,DEMIR Y,GUZELIS C.Moving towards in object recognition with deep learning for autonomous driving applications[C]∥International Symposium on Innovations in Intelligent Systems&Applications.[S.l.:s.n.],2016:1-5．
- [7]REN S,HE K,GIRSHICK R,et al.Faster R-CNN:towards real-time object detection with region proposal networks[J].IEEE Transactions on Pattern Analysis&Machine Intelligence,2015,39(6):1137-1149．
- [8]LIN T Y,GOYAL P,GIRSHICK R,et al.Focal loss for dense object detection[J].IEEE Transactions on Pattern Analysis&Machine Intelligence,2017(99):2999-3007．
- [9]ZHANG Y,SHEN Y L,ZHANG J.An improved tiny-yolov3pedestrian detection algorithm[J].Optik,2019,183:17-23．
- [10]HE K,ZHANG X,REN S,et al.Deep residual learning for image recognition[C]∥2016 IEEEConference on Computer Vision and Pattern Recognition.Las Vegas:[s.n.],2016:770-778．
- [11]RUSSAKOVSKY O,DENG J,SU H,et al.ImageNet large scale visual recognition challenge[J].International Journal of Computer Vision,2015,115(3):211-252．
- [12]EVERINGHAM M,WINN J.The PASCAL visual object classes challenge 2007(VOC2007)development kit[J].International Journal of Computer Vision,2006,111(1):98-136．
- [13]HE K,ZHANG X,REN S,et al.Delving deep into rectifiers:surpassing human-level performance on ImageNet classification[C]∥2015IEEE International Conference on Computer Vision,Santiago.[S.l.:s.n.],2015:1026-1034．
- [14]SZEGEDY C,LIU W,JIA Y,et al.Going deeper with convolutions[C]∥2015IEEE Conference on Computer Vision and Pattern Recognition,Boston:[s.n.],2015:1-9．
- [15]SZEGEDY C,VANHOUCKE V,IOFFE S,et al.Rethinking the inception architecture for computer vision[C]∥2016IEEE Conference on Computer Vision and Pattern Recognition.Las Vegas:[s.n.],2016:2818-2826.