机车驾驶路面状况监控系统障碍检测算法的研究  

Research on Roadblock Monitoring System of Locomotive Based on Video Analysis

作　　者：超木日力格[1] 赵守国[1] 王静[2] 刘硕研[2] 

机构地区：[1]北京交通大学计算机与信息技术学院,北京100044 [2]中国铁道科学研究院电子计算技术研究所,北京100044

出　　处：《铁路计算机应用》2012年第12期11-15,共5页Railway Computer Application

摘　　要：提出基于铁路智能视频监控图像序列分析的铁轨路障检测算法,即通过图像预处理突出铁轨信息,确定铁轨边缘像素;沿铁轨边缘建立检测窗,计算分析铁轨检测窗内的纹理特征;结合纹理特征及灰度特征进行铁轨障碍物的检测。在检测出路障目标的基础上对机车驾驶员提供实时的警报。实验结果表明,该算法检测速度快、可靠性高,可运用于实际路面状况检测系统。The paper proposed the railway roadblock detection algorithm based on the image order analysis of railway intelligent video monitoring, that was to stick out tile track information by image preprocessing, determine the edge pixel, set up the detection window, analyze the texture feature in track detection window, combine with the texture feature and gray feature to detect the roadblock. The report of roadblock detection was provided in real time for locomotive driver. The experiment result showed that the speed of detection was fast and reliability was high by using this algorithm. The algorithm could be used into practice.

关 键 词：机车驾驶 智能监控 障碍物检测 图像处理 

分 类 号：U268[机械工程—车辆工程] TP39[交通运输工程—载运工具运用工程]