Research And Application Of An FOA-based Edge Detection Algorithm For Images Of Mechanical Parts

With the rapid development of modern industry,the measuring precision of mechanical parts is more and more demanding,the traditional contact measurement method is not only inefficient,but also unfavorable to the automation and information of mechanical manufacturing process,in which the feedback control of each processing link can not be realized,so that a more effective measurement method is urgently needed.Based on the measurement technology of computer image processing among them,which has the advantages of high precision,high efficiency,non-contact,high degree of automation,and has a wide application prospect in the existing industrial environment.One of the important method to realize the non-destructive measurement is the high precision detection method of part image edge.Conventional Canny algorithms are too sensitive to noise and easy to detect isolated edge points and pseudo edge,as the double threshold method is used to locate the edge in threshold detection,the setting of high and low thresholds depend on prior knowledge and can not accurately locate all edges,the phenomenon of multiple responses at edge points is caused by noise,illumination and other factors in the actual detection,which makes the edge location inaccurate and the corner spot missed.In order to solve the problems of inaccurate location of edge points and missing detection of corner points,this paper using the Fruit Fly Optimization Algorithm(FOA),combined with the improved Canny operator and Hilbert transform,proposes an algorithm and implementation of edge detection of part image based on FOA.As a swarm intelligence algorithm,FOA has the advantages of fast convergence,few parameters,simple steps and no influence of internal factors,However,in the selection of parameter combination,algorithms tend to fall into local convergence without good experience.In this paper,the efficiency of the algorithm is greatly improved by optimizing the candidate solution mechanism,search radius mechanism and flight strategy mechanism.Firstly,using the Canny edge detection algorithm to extract edge points,the threshold is divided by linear combination of high and low thresholds to obtain prior knowledge of edge points;Then we use Hilbert transform to extract corner points,the edge tracking model based on Drosophila optimization algorithm is established by using edge points and corner points as heuristic information;Finally,the single pixel edge is obtained by correlation mechanism.Experimental simulation results show that under the condition of noise-free edge detection,compared with the traditional Canny edge detection algorithm,this algorithm has a higher accuracy;The part image is collected by the hardware platform of edge detection based on machine vision,a software Cognex(Connolly 2009)was used to verify the image edges.Experiment proved that the precision of part image detection is higher and has certain practical value.