Research Of Defects Recognition Method For Reflective Surface

In today’s era,the requirements for surface quality of products are getting higher and higher,and product surface defect detection technology is becoming more and more important in the production process.At present,the application of visual inspection technology is becoming more and more mature,and it has gradually become the mainstream of surface inspection technology.Based on this,this paper has carried out research work with the workpiece with reflective surface as the research object.Surface quality inspection is achieved by identifying reflective surface defects.In this paper,the refraction surface defect recognition technology at home and abroad is reviewed.On this basis,the method of reflective surface defect recognition based on visual inspection is studied.At the same time,this paper studies the techniques of surface image defect extraction,image stitching and surface defect classification and identification,and develops corresponding system software.Finally,experiments show that the proposed method can effectively extract and identify the defects of the reflective surface.The detailed research work is as follows:Firstly,this paper fully considers the noise,background texture and reflection phenomenon of the reflected surface image,and proposes an improved defect extraction algorithm based on Intrinsic Mode Function(IMF).By decomposing the image into a series of IMFs,the effects of image noise,surface texture and reflection on the defect extraction result are suppressed.Then the edge detection algorithm is used to detect the edge information of the image,and finally the complete defect information is extracted through the edge repair and image filling operations.Simulation results show that the method can effectively extract surface defects.Secondly,this paper studies the surface image stitching method and proposes an improved SIFT image stitching method.This method is based on the SIFT splicing method.The erroneous matching feature points are removed by establishing the angle of the line vector matching the feature points and the Euclidean distance,and the accurate registration information is calculated according to the matched matching feature points.Finally,the stitching of the image is done using multi-resolution fusion based on the best stitching.The simulation results show that this method can complete the seamless mosaic of images accurately and stably.Then,in order to identify the defects extracted from the reflective surface,the characteristic parameters of surface defects are studied in detail,and a defect recognition algorithm based on binary tree classifier is proposed.The defects are identified by key feature parameters.The system software of the experimental platform was developed,which can realize the functions of motion platform control,image acquisition,surface image mosaic,manual defect detection and automatic defect detection.Finally,the experimental research and analysis of the algorithm is carried out.Through the images collected by the experimental platform,the experiments of defect extraction algorithm,surface image mosaic algorithm and defect recognition algorithm are completed.The experimental results show that the proposed algorithm is feasible.The image mosaic algorithm can accurately and stably splicing the reflective surface image into a large view seamless image,and the image feature point matching accuracy rate is above 95%;the defect extraction and recognition algorithm can accurately identify the defects of the reflective surface.In addition,44 figures,4 tables,52 references are included in this paper.