Research On Roll Surface Defect Detection Technology Based On Image Mosaic

The roller is the main device on the rolling mill,and it is the tool to make the metal produce continuous plastic deformation.If the surface quality of the roller is not up to standard,the surface of the steel strip will have defects.Once the roll defect is not found or not dealt with in time,it may affect the quality of the product in the rolling process.Therefore,the factory attaches great importance to the inspection and maintenance of the replaced roll surface The traditional inspection and repair of roller surface defects are done manually.This method is affected by manual experience and subjective factors,and has low accuracy,high labor intensity and low efficiency.In order to realize the detection and recognition of roller surface defects and the subsequent repair of defects,the following research work is planned:(1)Image collection of roll surface.In order to obtain the image of the roll surface,the hardware design of the visual system is completed by selecting the industrial camera,lens and light source.After that,the surface of the roll is illuminated by light,so that the surface of the roll is clear and the brightness is moderate,and then the image of the roll surface is collected by an industrial camera.(2)Image preprocessing.In the process of image acquisition,the collected image will be affected by noise due to various factors,which is detrimental to the subsequent splicing,so it is necessary to preprocess the collected image.(3)Feature extraction.Aiming at the disadvantages of traditional SURF algorithm,such as slow extraction time and low precision of feature point extraction,optimization was carried out.Detection is proposed based on the strategy of chunking features,the image in front of the feature points extraction,first to block image,through the calculation of similarity of image block,divided into overlapping area,reduce the retrieval and calculation of the SUFR algorithm in the invalid region,narrowing the scope of matching,reduce feature matching algorithm so time,improve the efficiency of the whole algorithm of feature point detection and matching efficiency.(4)Image Mosaic.Finally,the SURF algorithm was used to match the image,and RANSAC algorithm was used to refine the matching points to improve the matching accuracy.The transformed image was obtained by affine transformation,and the images were aligned.Then,the image fusion algorithm was used to eliminate the stitching lines to achieve the image stitching and achieve the ideal stitching effect.Therefore,to solve these problems,machine vision is used to detect the surface defects of rolls,and the detection data are analyzed and the types of defects are identified,and then repaired by robots.Because the camera has a small Angle of view,it cannot capture the image of the whole roll surface.Experimental results show that the feature matching accuracy of SURF algorithm based on similar blocks is 80.4%,18.37% higher than that of traditional algorithm.Therefore,the quality of image Mosaic is improved,which is conducive to the detection of roller surface defects and the identification of defect types,and convenient for robot to repair roller defects.