Research On Circle Detection And Roundness Measurement Of 2D Component Measurement

In the two-dimensional measurement of components,it makes much sense to detect the circle quickly and accurately,and then calculate the roundness in the two-dimensional measurement.Circle detection algorithms are based on Hough transform algorithm,which can be divided into two parts.One of them is called probabilistic.The other is non-probabilistic algorithm.This paper improves the existing algorithms from the following aspects.Firstly,this paper uses three-point reverse to search connected subset in order to ensure gradual subset curve.Secondly,this paper calculates circles parameters based on the connected subset by using dichotomy method.A candidate circle can be determined when there are two similar parameters after finishing iterations with the use of dichotomy method.Thirdly,the true circle is confirmed by cumulating the edge point between tangent square and inscribed square.Fourthly,it comes to remove all connected subsets and the inner edge points of the circle.Fifthly,divide remaining edge points into regions.Successively,if the number of points in the region satisfies the conditions,this papers judges whether there is a circle by the random Hough transform algorithm.The circle detection algorithm in this paper improves the existing algorithm based on connected subset algorithm and random detection algorithm.Besides,the proposed algorithm takes various components of the measurement and processing into consideration.The results show that the proposed algorithm achieves a higher accuracy in a lower time complexity.Additionally,this paper proposes a multi circle detection algorithm based on the idea of multi-scale.The core of the idea of multi-scale is as follows.First of all,deal with image with the use of some effective methods.Then,extract the information of circles when the image has been transformed to a higher scale.After that,it comes to an end to calculate the number of circles and the location of rounds' edge rapidly and directly.This paper uses a relatively easier method to do conversion.Namely,in the condition that there are multi similar circular components,this paper calculates the first round.Then,iterations of images should be in the following according to the first round of regional image traversal.Finally,circles can be located by mapping the number of the region's edge points into the result on a scale of high level.After circle detection,we also need to calculate the roundness through the detected ring.This paper proposes an algorithm based on the exclusive balance principle of physics and the existing algorithms.We propose a method of roundness measurement based on the principle of mutual balance.The ring pixel will be treated as analog electronics in the simulated experiments.Firstly,place an initial point in the ring.Then,the point will move because of initial stress simulation by ring electronic group.The point will not stop move until the initial point find a comprehensive and balanced place where the pixel group repulsions ring.Finally,set the point as the center of circle.The experimental results show that the point has a better result than the least squares method.Moreover,the roundness measurement is suitable for the practical work.The paper is based on the research of 2D components measurement,which is focused on circle detection and roundness.We put forward a series of algorithms which have a good practical effect of the improved algorithm and optimization method.Besides,it is suitable for the real circle detection and roundness measurement work.In the end,it provides some new ideas for the future research.