摘要
为满足现阶段车间内无人驾驶车辆以及车辆自动导引时对车道线的识别速度和检测准确率的要求,课题组提出了一种结合LSD(line segment detect)直线检测以及ANN(artificial neural network)颜色判定神经网络的特定颜色车道线识别与拟合方法。实验表明:与传统的LSD直线检测相比,该方法在识别速度显著提升的同时,能够适应不同的检测环境,提升检测的准确率。
In view of the requirements of the recognition speed and detection accuracy of the lane lines for the unmanned vehicles and the automatic guidance of the vehicles in the workshop at this stage,a specific color lane line recognition and fitting method combining LSD(line segment detect)line detection and ANN(artificial neural network)color judgment neural network was proposed.The Experimental results show that compared to the traditional LSD linear detection,this method can not only improve the recognition speed significantly and adapt to different detection environments,but also improve the detection accuracy.
作者
陈燚
陈勇
王丙佳
邱洪斌
CHEN Yi;CHEN Yong;WANG Binjia;QIU Hongbin(College of Mechanical Engineering,Zhejiang University of Technology,Hangzhou 310014,China)
出处
《轻工机械》
CAS
2020年第5期62-66,共5页
Light Industry Machinery
基金
浙江省教育厅一般科研项目(Y201941882)。
关键词
直线检测
车道线识别
卷积神经网络
人工神经网络
straight line detection
lane line recognition
convolutional neural network
LSD(line segment detector)
ANN(artificial neural network)
