摘要
传统的盲区障碍物识别方法受到高速列车惯性的影响,导致识别结果不精准。为此,提出基于机器视觉的轨道车辆侧向盲区障碍物识别方法。在确定静态障碍物区域后,利用机器视觉方法分析不同目标轮廓,确定图像之间的变换关系,得到归一化图像坐标。计算不同图像之间的轮廓相似程度,完成静态障碍物识别。通过机器视觉的差分法提取出动态运动障碍物区域,计算立体置信度和平面置信度,由此识别动态运动障碍物。结果表明:该方法的灰度直方图检测结果与标准图像基本一致,证明其具有精准的识别效果。
To deal with the imprecise identification problem resulted from traditional blind area obstacle identification method affected by the inertia of high-speed train,an obstacle recognition transformation method based on machine vision is proposed.After the static obstacle area is determined,the machine vision method is used to analyze the contours of different targets and determine the relationship between images obtaining the normalized image coordinates.The contour similarity between different images is calculated to achieve static obstacle recognition.By the difference method of machine vision,the dynamic moving obstacle region is extracted,and the stereo and plane confidence are calculated to identify the dynamic moving obstacle.The results show that the gray histogram detection results of the proposed method are basically consistent with the standard image,which proves that the method has accurate recognition effect.
作者
王维华
WANG Weihua(Shaanxi Communication Technology College,Xi'an 710018,China)
出处
《机械制造与自动化》
2022年第6期212-214,230,共4页
Machine Building & Automation
基金
陕西交通职业技术学院2020年校极科研项目(YJ20002)。
关键词
机器视觉
轨道车辆
侧向盲区
障碍物识别
轮廓相似度
machine vision
rail vehicle
lateral blind area
obstacle identification
contour similarity
