Research On The Identification And Classification Of Complex Surface Defects On Aluminum Plate

In recent years,due to the rapid development of aerospace vehicles,automotive manufacturing industries,and high end electronic products worldwide,aluminum plate has a broad market prospect,as a high technology product application material and necessities for social development.The aluminum plate processing industry has developed rapidly in China.The output of aluminum plate production keeps growing,and the import and export trade volume has been increased for many years.However,the traditional aluminum plate processing industry is affected by backwardness of equipment,low level of production technology,and complicated production environment.After the production of aluminum plate,various complex defects will occur on the surface.For aluminum plates used in aerospace vehicles,automobile manufacturing industries and high-end electronic products,the complex surface defects will cause great losses to the aluminum plate industry.Therefore,what matters is that how to detect and identify complex defects in aluminum plate accurately and efficiently.In order to realize the online high-efficiency,high-precision detection of complex aluminum defects,this paper makes a research on the implementation of machine vision-based aluminum complex defect system and pattern recognition and classification algorithms.First of all,it explains the basic requirements and basic principles of the aluminum complex defect system,focusing on the basic architecture of the system;Then,the defect image is preprocessed witch is to reduce the background and remove the noise.The median filter is used to suppress the noise of the aluminum defect image,and the threshold adaptive method is used to segment the aluminum defect image.Next,according to the extraction of regional features,contour features,and grayscale features of several common aluminum target defects,the extracted defect feature data is standardized.Finally,for the four common defect types,this study uses the SVM classifier to identify and classify the tests,and a new identification mechanism is presented for possible new defects.By detecting new defect types,a new defect SVM sub-classifier is constructed.The new defect model is added to the common defect model to further improve the SVM classifier.The SVM classifier constructed in this paper has a good effect on the identification and classification of the four common types of defects,and we candiscover new defects,define new defects and establish new templates.The new mechanism combines SVM classifier with new defect classification mechanism?Not only does it improve the recognition efficiency of the common defects,but it can identify new defects and update the new defect creation template into the common defect templates.