Research On Visual Inspection Technology Of Automobile Fuse Box Quality

With the advancement of science and technology,the manufacturing industry is booming,and electronic components are gradually integrated.As the core of the car parts-the car fuse box integrates all the fuse strips in the car,while ensuring its performance,it also requires high efficiency and high precision in production.The quality inspection of automobile fuse boxes has also become the most important part of the automobile manufacturing industry,the traditional manual detection method has problems such as low detection efficiency and missed detection and false detection.,unable to meet the growing demand for high-volume inspection.On the other hand,the existing visual detection methods do not have a unified detection model,it is not possible to detect different types of products.Therefore,this paper studies the quality defect detection method of automobile fuse box,the main research contents are as follows:(1)Study the image segmentation algorithm,which integrates the BM3 D algorithm into the superpixel segmentation algorithm(SLIC),and replace the Euclidean distance with the Mahalanobis distance that closely combines the color distance with the spatial distance,while taking into account the area proportion of each scattered isolated pixel point in each surrounding clustering area,forming the B-SLIC algorithm,applying the improved algorithm to the segmentation of fuse box images,which can accurately divide multiple fuse strips in the fuse box.(2)In order to extract the position of the fuse blade in the fuse box,combined with the detection characteristics of the fuse strip,propose a method to find the maximum bounding rectangle of the fuse strip.This method utilizes the corner points of the fuse strip outline,fitting the outline of the fuse strip into a standard rectangle,obtain the workpiece reserved hole center as the reference point through Hough transform,and then obtain the coordinates of each fuse strip relative to the reference point,which forms the standard template information for the fuse box.(3)During the continuous detection process,the fuse box to be tested is often tilted,in order to use the standard template information to accurately extract the image of the fuse strip in the fuse box to be tested,studied the edge extraction algorithm of Canny image,use Fourier transform combined with Wiener filter instead of Gaussian filter to filter out image noise,and combined with a 3×3 window to calculate the neighborhood gradient magnitude.Extracting the edge of the fuse box with an improved algorithm,and then correct the tilt of the image through the tilt correction algorithm,which solves the problem of increased detection difficulty due to image tilt.(4)For different defects of the product,analyze its characteristics and design corresponding detection methods,through the gray histogram comparison to realize the missed insertion detection of the fuse strip;Complete the reverse insertion detection of the fuse strip through image matching;Design a "double check" method;which combines the two methods of color contrast and text recognition to form a double insurance,to detect the wrong insertion defect of the fuse strip,the experiment proves that the quality detection method of automobile fuse box proposed in this paper can quickly and effectively detect different kinds of quality defects.