Research On Machine Vision Identification Technology Of Three-core Power Line For Automatic Production And Testing

Three-core power cord is widely used in life and production.If the quality of the three-core power cord is not up to standard,the safety of life and property will be endangered in the process of power supply.At present,the production process of threecore power line in China is mainly completed by semi-automatic equipment.Many links need manual participation,low production efficiency.With the purpose of improving the production quality of the three-core power line and the means of machine vision technology,this paper focuses on the extraction of areas of interest,color recognition and edge feature extraction.The main work is as follows:(1)High quality image acquisition algorithm design: based on the principle of image color space conversion,the algorithm is designed to convert the YUV image data transmitted from the camera to the computer and decoded,and obtain the image data in the RGB color mode as the original image processed by the later image algorithm.(2)Image region of interest extraction algorithm design: an automatic global threshold segmentation algorithm is designed,a gray histogram model is constructed for the black background area,the gray value with the highest frequency in the histogram is automatically located,and the gaussian distribution is used to realize threshold segmentation.By using morphological operation,the image after automatic global threshold segmentation is denoised to obtain a complete region of interest,which lays a foundation for color recognition.(3)Color recognition algorithm design: in view of the color difference of the wire,the color classifier recognition method based on gaussian mixture model is designed,which can continuously train the color with color difference into the color classifier,so as to realize the color recognition conveniently and efficiently.(4)Edge feature extraction algorithm design: in order to solve the problem of deformation and unfixed position of wire after riveting,a template search method based on square error matching is designed,and the riveting area is extracted by affine transformation.In the riveted area,the edge features of the bare copper wire outside the terminal insert are extracted by using the band-pass filter algorithm to realize defect detection.(5)Machine visual detection system hardware structure and software system design: according to the characteristics of automatic production and detection of threecore power line,a suitable optical path was built to collect high-quality images,and the detection software architecture,process and human-computer interaction interface were designed according to the detection requirements.Multiple experiments were designed to verify the above image processing algorithm,and the identified wrong image was analyzed in detail,and the cause of the error was given.