The Research On Localization Algorithms Of Mark Points Based On Machine Vision Technology

Mark points positioning is a common kind of location methods in the machine vision detection and location technology. Especially,it is used in the process of OGS touch panel,chips、PCB-board and HDI in a large scale. But in the process of these high precision machining,Mark points positioning accuracy may fall because of the flaw or plaque noise on the edge of Mark points. In this paper,an algorithm is proposed is aimed at removing defects edge to improve the accuracy of Mark points positioning.This research converts the original image to(ρ,θ) domain making use of Radon transform. And the gray scale and shape will change if the original image has defects edges. So we can remove defects edges by screening gray scale and distance in order to improve the accuracy of the results which least squares method calculates.Firstly,standard template of(ρ,θ) domain can be calculated by Radon transform when the original image has no defects. As for circular Mark points,it compares two brightest points on the upper part and lower part with those in the standard template for gray scale screening. Because the width of circle in(ρ,θ) domain is constant,and we can furtherly remove defects edges by screening the width. Finally,center coordinates can be calculated by making use of least squares method.As for rectangular mark points,the points of the original picture in(ρ,θ) domain which is calculated in Radon transform will be remained if their grayscale exceeds the threshold. And the distance of two bright points in the same angle represents the length or width of the rectangle,so it can furtherly remove defects edge by screening the width.Finally,it can find out the four lines which the edges of the rectangle lie in.And center coordinates can be calculated by the four vertex coordinates which is calculated by the four lines above.Ultimately,the experiments show:If the imperfect circular Mark points have partly perfect corresponding arcs,this algorithm can remove defects edges to improve the accuracy.And if the rectangular mark points is imperfect,this algorithm can also remove defects edges to improve the accuracy.