Research On Surface Defect Detection Algorithm Of Automotive Aluminum Die Castings Based On Image Processing Technology

As the best lightweight material,aluminum alloy casting plays an important role in industrial manufacturing.During the processing and transportation of aluminum castings,due to some uncertain factors,defects such as pores,cracks,scratches and black spots will inevitably occur on the surface of aluminum castings,which will affect the appearance and quality of products.Therefore,the production department also puts forward higher and higher requirements for the detection method and detection accuracy of casting surface defects.At present,the product surface defect detection technology has developed from the initial manual visual detection technology to the machine vision automatic detection technology represented by CCD camera and image processing technology.Among them,the traditional manual visual inspection method has the problems of low efficiency,easy fatigue and poor real-time performance,which can not meet the needs of the production line.Image processing and pattern recognition based on CCD camera have become a new detection method in the field of industrial detection.However,at present,the automatic detection technology for products with poor image quality,complex surface structure and different types of defects（such as surface defect detection of aluminum castings）still needs further research.Therefore,combined with the surface defect detection of aluminum alloy castings,this paper studies the algorithm in the detection of automobile front subframe,The new surface defect detection method of aluminum castings with complex structure is verified and analyzed,which lays a good foundation for ensuring the quality of industrial products.The main research contents of this paper are as follows:（1）Research on surface defect segmentation algorithm.A surface defect segmentation algorithm based on improved level set collaborative spatial fuzzy clustering is adopted.This paper is due to the characteristics of complex surface structure and different defects shape to automobile subframe.Firstly,the part image is preprocessed by wavelet neighborhood shrinkage denoising and multi-scale enhancement algorithm,and then the defect region is preliminarily located by level set function.Finally,combined with spatial fuzzy clustering,the defect region is further accurately located to achieve the purpose of surface defect segmentation of aluminum castings.Through experiments,four kinds of defects such as surface crack,indentation,color spot and hole of aluminum casting are detected.Compared with other algorithms,it can achieve better results in effect,accuracy and recall,lay a good foundation for subsequent surface defect classification,and can meet the requirements of industrial detection.（2）Research on classification method of surface defects.Two classification and recognition methods,namely "one-to-one" method and "one-to-one" method,are designed to classify and identify casting surface defects respectively.Extract internal features（texture information,grayscale information）and external features（geometry information）for the segmented defect area,and the PCA operator is used to reduce the dimension of the relevant feature set to construct a new feature data set.Subsequently,combined with the two classifier models designed in this chapter,the surface defects of three types（cracks,color spots and holes）of aluminum castings are classified and identified respectively.The experimental results show that the accuracy of this method is 94.1%,which is better than other classification methods.（3）Research on SVM kernel parameter optimization algorithm.In order to further improve the accuracy of surface defect classification of aluminum castings,this paper proposes a gray wolf optimization algorithm（FGWO）based on Fuch chaos strategy and nonlinear convergence factor.This method introduces Fuch chaos reverse learning strategy,dynamic nonlinear control parameters,dual weight factor strategy and survival of the fittest selection strategy,respectively in the three stages of the algorithm,The purpose is to provide a new mechanism for balancing the performance of global exploration and local development,and to enhance the convergence speed and accuracy of the algorithm.In order to verify the performance of the algorithm,based on 10 standard test functions,fgwo algorithm is compared with other algorithms.Experimental results show that fgwo algorithm can effectively solve the function optimization problem and is better than other algorithms.Finally,in the SVM classification problem,Taking fgwo algorithm as a new strategy and applied to the parameter optimization process of support vector machine,a fgwo-svm classification model is constructed to realize the classification of surface defects of aluminum castings.The experimental results show that the average recognition rate of the model is better than the classifier model proposed in Chapter 3,and the accuracy of the whole model is improved by 4%.