Research On Detection And Recognition Method Of Steel Pipe Surface Defects Based On Machine Vision

Surface quality is an important indicator for evaluating the quality of steel pipes,and it has a crucial impact on the performance and quality of products.At present,there are few researches on machine vision inspection of steel pipes at home and abroad,and there are no mature products.This paper takes hot-rolled seamless steel tubes as the research object,designs and builds an experimental platform for collecting images of surface defects on steel tubes,acquires and constructs image samples of surface defects,and conducts detection and identification classification studies,as follows:(1)In order to obtain images of steel pipe surface defects,a method for collecting steel pipe surface images is proposed,which can adapt to steel pipes with a certain range of outer diameters.The bright field illumination method is used to design the imaging optical path diagram of the steel pipe surface,and the imaging hardware parameter design is based on this.Set image acquisition parameters to obtain and build a sample library of steel tube surface defects images,which can be used as the source of image data and performance evaluation basis for detection and identification classification methods.(2)Based on the image sample library,research and development of steel tube surface defect detection method.Aiming at the problem that the curved outer surface of the steel tube is likely to cause uneven illumination,a detection method based on the improved Kmeans grayscale forward and inverse summation is proposed.First,the vertical projection method is used to obtain the image of the steel pipe area,and the gray-scale inversed image is calculated.The FMR algorithm is used to enhance the steel pipe area image and the grayscale inversed image to obtain uniform and high-contrast images of their respective backgrounds.-means segmentation algorithm processing,obtain the respective defect results,and sum the two results,and finally perform image post-processing to optimize the results,and locate and extract the defect area.(3)On the basis of detection,according to the characteristics of the defect area,a classifier for identifying defects on the surface of the steel pipe is designed.Extract the shape features,grayscale features and texture features of the defect area,analyze the difference of the normalized feature values,select the effective features as the input of the support vector machine classification algorithm,and optimize the important parameters of the support vector machine algorithm to construct the classifier For defect identification classification test.(4)By constructing an image sample library,an experimental study was conducted to test the performance of the steel tube surface defect detection method and defect recognition classifier.The results show that the method in this paper can effectively detect the defects of pits,warping,scratches and roll marks on the surface of the steel pipe,and has strong anti-interference ability to the change of illumination.The above four defects have good recognition performance,and the comprehensive recognition rate reaches 91.25%,which is in line with the expected effect.