| With the continuous development of the automobile industry,the coating process of the whole body and its parts has also been developed rapidly.In order to ensure the good appearance and long service life of its parts,it is necessary to carry out strict quality inspection on the painted auto parts.Ensure that the specified quality level is achieved.Defects on the surface of painted auto parts mainly include scratches,bumps,gluing,sagging,etc.The location and size of defects appear randomly,and the detection accuracy is high.Traditional manual detection of defects in automotive painted parts has efficiency and detection accuracy low and high labor costs.With the continuous development of machine vision technology,the application fields are becoming more and more extensive.Using machine vision detection to replace manual detection can not only reduce production costs,but also further improve detection efficiency and detection accuracy.According to this,this paper designs a coating defect detection system based on machine vision,researches and analyzes the defect characteristics on the surface of painted automobile parts,and realizes the research and use of related detection algorithms in Halcon MATLAB.The main research contents are as follows:1)Build the hardware platform of the detection system to obtain the corresponding image information.It mainly includes the design of the mechanical transmission structure and the comparison and selection of the lens,camera,light source and other devices in the image acquisition system.2)An image preprocessing scheme for two types of problems of image background noise removal and image illumination unevenness is studied.Firstly,the causes of image background noise and image illumination unevenness are analyzed at the hardware level and the principle is introduced,and corresponding solutions are proposed for the removal of background noise and the different characteristics of balanced illumination.The background noise is removed by constructing different kinds of low-pass filters in the frequency domain,and an image preprocessing method based on homomorphic filtering is proposed for the uneven illumination of the image,which realizes the pre-proposed image preprocessing effect.3)Research on image segmentation and defect extraction algorithm of region of interest.Different algorithms of region of interest segmentation are introduced,and an algorithm of sub-pixel edge extraction and automatic threshold segmentation is proposed to segment the region of interest.When the surface defect extraction algorithm is studied,the corresponding defect detection algorithm is proposed according to its different defect characteristics.A gray-scale variation defect detection algorithm based on the principle of gray-scale projection,a defect extraction algorithm based on morphological Blob and an optimization method based on the watershed algorithm are proposed,and these algorithms are practically applied to verify the correctness of the algorithms and reliability.4)Set up the human-computer interaction interface of the upper computer.The interactive software of the human-machine interface consists of three parts,which are respectively composed of the login interface,the.NET vision system framework based on C# and the parts based on the Halcon defect detection algorithm nested in the framework.On the basis of realizing the detection function,the system has good operability.This system builds an experimental platform and conducts actual experimental tests.The test results show that this system can quickly and accurately identify and extract the characteristics of coating defects.The average single recognition time is320 ms,and the recognition accuracy is 97%.Compared with manual detection efficiency Increase by nearly 40%,meet the requirements of industrial coating defect detection. |
PDF Full Text Request