摘要
该自主车辆地面自动辨识技术是以Leobot-Edu自主车辆作为试验载体,并应用DH-HV2003 UC-T视觉传感器对常见的5种行车路面(石子路面、水泥路面、土壤路面、草地路面、砖地路面)进行图像信息的采集,应用Matlab图像处理模块对其依次进行压缩编码、复原重建、平滑、锐化、增强、特征提取等相关处理后,再应用Matlab BP神经网络模块进行模式识别。通过对模式识别结果分析可知,网络训练目标的函数误差为20%,该系统路面识别率达到预定要求,可以在智能车辆或移动机器人等相关领域普及使用。
The ground automatic identification technology for intelligent vehicle is taking Leobot-Edu autonomous vehicle as a test vector and using DH-HV2003UC-T vision sensor to collect image information of five common lane roads（cobbled road, concrete road, dirt road, grass road, tile road）, then using MATLAB image processing module to perform coding com- pression, recovery reconstruction, smoothing, sharpening, enhancement, feature extraction and other related processing, then using MATLAB BP neural network module to carry on pattern recognition. Through analyzing the pattern recognition result, it shows that the objective error is 20%, the road recognition rate has reached the intended requirement in the system, and it can be universally applied in the smart vehicle or robots and other related fields.
出处
《现代电子技术》
2011年第10期8-11,16,共5页
Modern Electronics Technique
基金
河北省自然科学基金资助项目(E2008000098)
关键词
自主车辆
视觉传感器
图像处理
模式识别
intelligent vehicle
vision sensor
image processing
pattern recognition
