基于亚像素边缘检测的板材直径测量研究

Research on plate diameter measurement based on sub-pixel edge detection

因像素级边缘检测精度低,导致板材的测量存在较大的误差。为此,提出了一种基于亚像素边缘检测的板材高精度测量方法。该方法假设物体边缘穿过的像素单元灰度值是不连续的,在各个边缘像素单元的矩形邻域内建立模型,准确计算出边缘的亚像素位置。通过张正友标定法校准相机,将图像坐标系与世界坐标系建立联系,计算出圆形PVC板材的直径。实验结果表明,在合成圆位置检测方面,半径的均方误差为0.034 1,检测精度较高。在实际图像测量方面,板材直径的均方误差为0.018 6,验证了该方法具有较高的稳定性和精度。

Due to the low accuracy of pixel-level edge detection,there is a large error in the measurement of the plate.Therefore,a high precision measurement method of sheet metal based on sub-pixel edge detection is proposed.This method assumes that the gray value of the pixel unit through which the object edge passes is discontinuous,and establishes a model in the rectangular neighborhood of each edge pixel unit to accurately calculate the sub-pixel position of the edge.The camera is calibrated by Zhang Zhengyou calibration method,the image coordinate system is connected with the world coordinate system,and the diameter of circular PVC sheet is calculated.The experimental results show that the mean square error of radius is 0.0341,and the detection accuracy is high.In the actual image measurement,the mean square error of the plate diameter is 0.0186,which verifies that the method has high stability and accuracy.