Studies Of Ellipse Detection Algorithm In Digital Image Processing

Ellipse detection is an important technology in digital image processing and computer vision,and it has a wide range of applications in the field of industrial automation monitoring,biomedicine,aerospace and so on.Therefore,it is very important and meaningful to develop a fast and accurate ellipse detection algorithm.This paper begins with a brief introduction of the research status of ellipse detection algorithm at home and abroad,and then deeply studies three kinds of ellipse detection algorithms.The ellipse detection algorithm based on direct least squares method is simple and clear,but the algorithm program is running for a long time,and it is easy to make a false fit for the non-elliptical contour.Namely,the algorithm is affected largely by interference points.The ellipse detection algorithm based on Hough Transform in one dimensional parameter space is a novel algorithm,which can reduce the parameter space of the Hough Transform to one dimension,but it depends on the end points of the major axes of the ellipse.If the end points of the major axes aren't available,the algorithm will fail to detect ellipses.In view of the disadvantages of the ellipse detection algorithm based on Hough Transform in one dimensional parameter space,this paper improves the algorithm,which makes the algorithm still valid when the end points of the minor axes aren available.Then,this paper presents a new algorithm for ellipse detection based on graph symmetry,which adopt the method of calculating the parameters step by step in order to compute the parameters of the ellipse step by step,so as to reduce the overall time complexity of the algorithm.The algorithm calculates the inclination angle of an ellipse by using the symmetry of the ellipse,after finding the center of the ellipse,and then further determine other parameters of the ellipse.In addition,the algorithm's method to calculate the parameters of ellipse is simple,not also depending on the end points of the major or minor axes of the ellipse,but the outline of the ellipse must be more than 1/2,otherwise the algorithm will be invalid.In this paper,the three ellipse detection algorithms are all tested respectively to verify the effect of their ellipse detection.Finally,three kinds of ellipse detection algorithms are applied to the automatic counting of red blood cells.Experiments are carried out and the advantages and disadvantages of the three algorithms are compared.The experimental results show that,under the condition of no adhesion of red blood cells,the three algorithms can obtain a good counting effect.Then,we also analyze the two problems in the use of the ellipse detection algorithm for red blood cell counting,namely:(1)the influence of white blood cells;(2)the effect of cell adhesion.In view of these two questions,this article has carried out the detailed explanation and has given the relative effective solutions.