Digital Processing And Intelligent Recognition Of Welding Surface Defect Image Of Amusement Structure

Many major parts of amusement facilities,such as track,big arm and safety bar,are mostly connected by welding.Welding not only accounts for more than 50% of the workload of the whole amusement equipment manufacturing link,but also is the most important link in the amusement equipment manufacturing link.Since amusement facilities are typical equipment affected by dynamic load,it can be said that the quality of welding determines the quality of amusement facilities and is also the premise of safe operation of amusement facilities.With the development of The Times,more and more people began to pursue more exciting forms of entertainment,which also makes the structure and operation of amusement facilities began to develop in a complicated direction,so the welding quality more need to be strictly controlled.Amusement facilities welding parts in front of the nondestructive inspection surface testing is required,but this link is often completed by artificial,there is a low efficiency,large amount of work tasks,the problem such as subjective factors influence the results,Therefore,this article designed a set of amusement facilities Welding surface defect identification system assists manual appearance inspection to improve the intelligence and standardization of welding defect appearance inspection.The main research contents of this article are as follows:In view of the three defects of surface porosity,surface crack and edge bite in the welding process of amusement facilities,a mobile phone was used to take photos for collection.The collected welding defect images were imported into Matlab software for image processing.Firstly,the welding defect images were preprocessed.This process included the use of weighted average method to conduct grayscale processing on RGB images to enhance image contrast and highlight image details.Gaussian homomorphic filter is used to eliminate the effect of uneven illumination.The median filtering method was used to eliminate salt and pepper noise.The image is segmented by Otsu algorithm to distinguish the background from the target area and highlight the target contour.Secondly,the defect characteristic parameters of the welding defect image were extracted.In this process,canny edge detection method was used to mark the points with obvious brightness changes in the image.The expansion algorithm is used to fill the void in the target region and eliminate the noise of small particles contained in the target region.By means of image small area elimination and target defect graphic marking,the defect area of welding is accurately extracted.Finally in the Matlab software to extract a variety of characteristic parameters,and choose the Area,MajorAxisLength,MinorAxisLength,Eccentricity,EquivDiameter,Solidity,among other,generating,ThinnessRatio,AspectRatio 10 kinds of parameters as characteristic parameters of welding flaws in the image.In the study of welding surface defect identification method of amusement equipment,the model,relation and design process of BP neural network are introduced in detail.Secondly,by comparing the different hidden layer and hidden layer neurons on the influence of the BP neural network,through the test to determine the recognition efficiency of the best number of hidden layer and hidden layer neurons,the use of Matlab and Python two languages mixed programming,establishing rides welding surface defect recognition system,and the practical application of this system are introduced.