Research On Defects On The Metal Surface Based On Image Processing

With the development of computer technology, the advantages of defects detectionof products surface based on image processing is more and more obvious. Metalsurfaces is highly reflective and drawing, therefore, the defect detection of metal surfacehas been a very difficult problem.In this paper the dirt and scratches of the metal surfacewill be treated as the object of study. Through the analysis and research of a largenumber of metal images, a series of effective specific theory and algorithms to identifyand quantify dirt and scratches on the metal surface is formed.First, in order to extract the area of dirt and scratches on the metal surface moreaccurately, the images of the metal need to be preprocessed. An improved imageenhancement method based on wavelet is proposed. It can effectively identify the targetarea of the image. The experimental results show that, the algorithm not only has thedesired effect of image enhancement, but also can better remove the noise and preservethe image detail. Secondly, through the analysis and comparison of the characteristicsand scope of various commonly used threshold segmentation algorithm, according tothe characteristics of color and brightness of metal defects, the most appropriate methodfor image segmentation is chosen. Finally, in the process of quantifying, thesegmentation image is filtered using the improved method of morphological filtering inthis paper. Experimental results show that the method can not change the image size,atthe same time, it can effectively filter out the interference isolated nodes and glitchnoise on the boundary, it can also fill the loopholes in image area.Then the simple andeffective method of pixel count is used to calculate the area and length of the defect.In this paper, the dirt and scratches on the metal surface are identified andquantified. It lays the foundation for establishing a complete set of defect detectionsystem of metal parts based on image processing and pattern recognition, it is highlysignificant for improving the production efficiency of metal parts.