Research On On-line Detection Method Of Oil Seal Defect Based On Image Processing

With the continuous improvement of industrial production automation,it becomes more and more important to realize the mass production and automatic detection of products.Based on the current production status of the oil seal industry,this subject proposes an online detection method for oil seal defects based on image processing in view of the shortcomings such as low detection efficiency,easy omission,and high cost when employees perform visual inspection of oil seal defects.The oil seal image is collected through the image acquisition system.The oil seal image is pre-processed,including gray processing,median filtering,and then the Otsu threshold segmentation and chain code method are used to extract the contour features of the oil seal.According to the feature value information of the detection result,there will be Defective samples are screened.Experimental results show that the detection system can complete product defect detection online.The method has the advantages of low detection cost,high reliability,good real-time performance,and easy online implementation.On the basis of the existing online inspection methods of oil seal defects,further research was conducted by consulting relevant data,and a reasonable inspection scheme was designed and a reasonable inspection scheme was designed.Combined with the production requirements of oil seal enterprises,the overall scheme design of the online inspection system for oil seal defects,including detailed selection of hardware selection.Combined with the theoretical selection of image acquisition CCD camera and optical lens,installing the camera and lens,connecting the camera to the host computer,analyzing the characteristics of several common lighting sources,using a reasonable ring LED light source.Designed and built an image acquisition experiment platform,and selected other hardware related to this system.Several methods of image filtering and image segmentation were analyzed and compared.Based on the characteristics of the experimental image and the respective advantages of various methods,median filtering and threshold segmentation were selected to process the oil seal image.This paper analyzes the edge detection of oil seal images,introduces several classic detection operators and edge detection algorithms based on wavelet transform,and proposes an edge detection algorithm based on chain code method for oil seal edge extraction.Classify several kinds of defects of oil seal defects,analyze and compare the respective characteristics of several classification theories,including K nearest neighbor classificationalgorithm,decision tree classification algorithm,neural network method and support vector machine method.Based on the types and characteristics of oil seal defects in this subject,a1-1 discrimination method based on SVM was selected to construct an SVM classifier for oil seal defect classification.The experiment proves that the system can realize the oil seal defect detection,and the detection accuracy can reach 0.01 mm,which meets the requirements of the system technical indicators.