摘要
为实现工程机械实际行走速度的实时在线检测,通过建立一种基于机器视觉的间接测速模型,把摄像机安装在行走的工程机械车辆上,并对摄像机进行标定确定出摄像机内外参数;在工程机械实际行走过程中实时采集前方路面的连续帧图像进行光流分析,根据运动主方向原理来建立光流运动方向;实验采用一种基于改进的金字塔式Lucas-Kanade光流算法计算出路面连续帧图像的光流速度;统计出所有图像间的平均光流速度后,根据标定所得的摄像机参数、图像坐标与世界坐标的关系计算出实际行驶速度。测试结果与实际速度比较得出误差为1.45%,结果表明该方法能精确反映工程机械的实际行走速度。
In order to realize the realtime and online speed detection of the traveling construction machinery, an indirect speed detection model was established by fixed the camera on the traveling construction machinery, and the camera was calibrated to obtain the interior and exterior parame ters. The camera collected the sequential frame images of the road in front of the traveling vehicle for the optical flow analysis. The main migration direction theory was adopted to analize the moving di rection. The optical flow speed of interframe was ealuculated by using improved Pyramidal Lucas Kanade optical flow algorithm. After determining average optical flow speed between all images, the actual travel speed was estimated according to the relationship between the image coordinate of the camera and the world coordinate based on camera calibration parameters. By comparing the test re sults and the practical speed, the error was found to be 1.45% , suggesting that this model can re fleet the actual speed of construction machinery correctly.
出处
《广西大学学报（自然科学版）》
CAS
CSCD
北大核心
2012年第5期907-912,共6页
Journal of Guangxi University（Natural Science Edition）
基金
教育部中央高校基本科研业务费专项资金项目资助(CHD2010ZY011)
关键词
机器视觉
工程机械
光流
速度检测
machine vision
construction machinery
optical flow
speed detection
