基于改进Canny边缘检测算法的道路标线检测实验研究  

Research on experiment of road marking detection based on improved Canny edge detection algorithm

作　　者：李志鹏 于鸿彬[1,2] 邵宏宇[3] LI Zhipeng;YU Hongbin;SHAO Hongyu(School of Mechanical Engineering, Tianjin Polytechnic University, Tianjin 300387, China;Tianjin Key Laboratory of Modern Mechanical and Electrical Equipment Technology, Tianjin Polytechnic University, Tianjin 300387, China;School of Mechanical Engineering, Tianjin University, Tianjin 300072, China)

机构地区：[1]天津工业大学机械工程学院,天津300387 [2]天津工业大学天津市现代机电装备技术重点实验室,天津300387 [3]天津大学机械工程学院,天津300072

出　　处：《实验技术与管理》2019年第9期137-141,148,共6页Experimental Technology and Management

基　　金：国家重点研发计划项目(2 0 1 7 Y F B 110 4 20 2,2016YFB1102003)

摘　　要：为提高传统Canny算法的灵活度和自适应能力,提出了一种改进的自适应Canny边缘检测算法。采用双边滤波代替高斯滤波滤除噪声并锐化图像边缘;再使用水平、垂直、45°、135°等4个方向的梯度模板对图像的梯度幅值进行计算。对传统的Otsu阈值分割算法进行了改进,改进的思路是找出类内和类间方差的最大值作为阈值,阈值搜索范围的缩小可以使计算量减少,实现快速分割。通过道路标线图像验证,说明改进后的Canny算法对道路标线的分割效果更好,减少了边缘断裂和假边缘,处理的时间也相对缩短。To improve the flexibility and adaptability of traditional Canny algorithm, an improved adaptive Canny edge detection algorithm is proposed. Bilateral filtering is used instead of Gauss filtering to remove noise and sharpen image edges. Then gradient magnitude of image is calculated using gradient templates in four directions: horizontal, vertical, 45° and 135°. The traditional Otsu threshold segmentation algorithm is improved, and the idea of improvement is to find the maximum value of intra-class and inter-class variance as the threshold. Reducing the search range of threshold can reduce the amount of calculation and achieve fast segmentation. The validation of road marking image shows that the improved Canny algorithm can segment road marking better, reduce the edge breakage and false edge, and shorten the processing time relatively.

关 键 词：CANNY边缘检测 道路标线检测 双边滤波 梯度幅值 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]