Study And Development Of Surface Defect Detecting System Of Stamping Parts Based On Image Processing

With the improvement of the press line automation requirements, automatic detection of products has become an important indicator to the degree of automation. However, the detection of stamping products is still in the manual detection stage for now. Low efficiency, bad stability, high cost of manual detection process seriously restricted the automation process of stamping production line. As a new type of NDT technology, machine vision inspection technology is gradually promoted in the detecting field of industrial production with its efficient, high-speed, stable, accurate, flexible features which also provides the new ideas for the detection of stamping products surface defects. However, at present, machine vision inspection technology in the field of stamping products surface defect detection is generally limited to the stamping products with single large planar surface. It is not widely used for the stamping products with non-planar surface and complex features surface. Therefore, in this article, the cambered stamping part is chosen as the research object, a surface defects inspection system of stamping product based on machine vision has been developed.According to the inspection requirements and the surface defects of the non-planar stamping products, a serious study had been conducted to achieve the design of the machine vision detection system:1. In order to meet the requirements of detection accuracy, the description of defects characterization and their reasonable classification had been explicated by analyzing their formation mechanism and morphology. Detection indicators of the inspection system had been explicated, and the surface defects of the cambered stamping part had been divided into three categories:rupture, scratch, indentation. The overall structure of the machine vision inspection system had been designed according to detection content and indicators. To meet the detection performance requirements, the selection of camera, optical lens selection and light source had been completed. The related parameter of the inspection system had been checked and the theoretical detection accuracy achieves 0.15mm.2.For the problem of uneven illumination and low contrast, the best lighting solution of the defects had been designed according to the analysis of defects’morphology and light-reflecting properties3.For the problems of defects segmentation, image processing algorithm of each defects based on HALCON had been designed by analyzing defects image. In the rupture detection process, in order to exclude interference of the structure shadow, an image segmentation method based on standard image subtraction combine with grayscale matching and construction of interactive ROI is proposed. This method can exclude the interference stably, and segments out the rupture region completely. For scratch detection, a local dynamic thresholding method is proposed by comparing with the traditional segmentation method after the unsharp masking enhancement. It is proved that this method can achieves a more stable segmentation.4. For the defects classification, an automatic defect classification method is proposed based on multi-shape feature which chooses the roundness, tightness, eccentricity, and the length of Euclid as the basis for defects classification.5. For the shortcomings of traditional measurement methods to defect features, the single-camera vision measurement technology has been discussed. The transformation between the image coordinate and the three-dimensional coordinates of the world can be calculated quickly by using camera calibration method based on HALCON. It can be used to achieve automatic measurement of defect features. It is proved that this method can achieve automatic measurement to stamping surface defects features accurately and quickly.6. The framework as well as the detection process of the software system had been designed according detection algorithm. Man-machine friendly interface of the software system had been designed by using the HALCON combine with VC++ which contained debugging, testing and defect information saving functions of each detection items. The results show that this method can improve the efficiency of system development and shorten the development cycle.The testing results show that:stability and processing speed of the system requirements are met. Innovation of this article is to propose a machine vision detection solution to the non-planar Stamping products which can provide the new ideas for the application of machine vision detection technology on defect detection of complex stamping products.