基于粗糙性度量的道路可行驶区域识别方法  

Roughness Measure-Based Road Drivable Region Identification

作　　者：马雷[1] 刘泽宾 曹彪 王文举 邱泉源 Ma Lei;Liu Zebin;Cao Biao;Wang Wenju;Qiu Quanyuan(College of Vehicle and Energy,Yanshan University,Qinhuangdao 066004)

机构地区：[1]燕山大学车辆与能源学院,秦皇岛066004

出　　处：《汽车工程》2021年第11期1594-1601,1610,共9页Automotive Engineering

摘　　要：本文中提出了一种基于粗糙性度量的道路图像可行驶区域识别方法。首先定义道路图像的粗糙度信息,获得道路图像色彩分布直方图及其上下近似信息,通过粗糙性度量实现了道路图像初分割;然后依据初分割图像色彩区域色差和像素规模进行区域合并,获得道路可行驶区域特征;再利用改进的区域生长算法对特征图像进行识别;实现了对道路可行驶区域的识别。实验结果表明,该方法取得了良好的识别效果。A method for identifying the drivable region of road image based on roughness measure is pro⁃posed in this paper.Firstly,the roughness information of the road image is defined to obtain the color distribution histogram of the road image and its upper and lower approximation information,and the initial segmentation of the road image is realized through the roughness measurement.Then,region merger is conducted according to the color difference and pixel scale in the color regions of the initial segmentation image,to obtain the characteristics of the drivable regions of the road.Next,the improved region growing algorithm is used to identify the feature image.Final⁃ly,the recognition of the drivable regions of the road is realized.The results of experiment show that the identifica⁃tion method has achieved good results.

关 键 词：道路图像 粗糙集 区域生长算法 K-MEANS聚类 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]