摘要
基于双目立体视觉模型,设计了一种工件尺寸测量系统。该系统通过两个工业数字摄像头实时采集工件不同角度的两幅图像,并基于机器视觉软件Halcon的HDevelop开发环境,对两幅图像进行相关预处理,获取感兴趣区域;提出了一种基于Canny的亚像素边缘检测从而获取精确的边缘特征点,并通过双目系统标定、立体校正、极线约束与NCC相结合的立体匹配,通过坐标系转化再结合几何方法,以此来测量工件尺寸。通过大量实验证明:该方法可测得的工件参数和实际情况没有较大的偏差,检测精度达±0.3 mm以内,且简单快捷,具有较好的准确性和实时性。
Based on binocular Stereo Vision Model,a dimension measurement system for workpiece is designed in this paper.The system collects two images of the workpiece from different angles in real-time by two industrial digital cameras,and gets the region of interest by pre-processing the two images based on the HDevelop environment of Halcon.In this paper,a method based on Canny sub-pixel edge detection is proposed to obtain accurate edge feature points,which is combined with NCC through binocular calibration,stereo correction,epipolar constraint and coordinate transformation to measure the size of the workpiece.Through a lot of experiments,it is proved that there is no big deviation between measured parameters of the workpiece and actual situation.The precision of this method is within±0.3 mm,which is simple and fast,and has better accuracy and real-time.
作者
房家军
郁汉琪
花慧敏
Fang Jiajun;Yu Hanqi;Hua Huimin(Nanjing Institute of Technology,Nanjing,Jiangsu 211167,China)
出处
《绿色科技》
2020年第18期219-223,共5页
Journal of Green Science and Technology
关键词
双目立体视觉
摄像机标定
边缘检测
特征点提取
立体匹配
binocular stereo vision
camera calibration
edge detection
feature point extraction
stereo matching
