Research On Surface Defect Detection Of Mobile Phone Shell Based On Camera RGB Channels Image Fusion

Aiming at the characteristics of various types and optical properties of mobile phone shell surface defects,the imaging optimization and defect fusion detection methods of mobile phone shell surface defects based on machine vision are studied in this thesis.Taking the aluminum alloy mobile phone shell as the specific research object,analyzed the characteristics and optical properties of surface defects,and proposed the imaging method of "using a color camera to collect images under multiple light paths" and the defect recognition method of "RGB three-channel image separately processing corresponding defects and then fusing the results".The imaging effect is equivalent to the traditional method of " each light path matches a camera under multiple light paths ",moreover,it can further design special algorithm for single channel image defects to effectively improve algorithm efficiency and pertinence,and the hardware system has low cost and high flexibility.It provides solutions to the surface defect detection of other mobile phone shells with different materials and colors.The main research contents of this thesis are as follows:(1)An imaging method of RGB three channels corresponding to three different light paths for a color camera is presented.The three imaging light paths adopt LED stripe red parallel light source and green parallel light source and the blue coaxial light source respectively.Summarize the characteristics of typical defects such as scratches and spots on the surface of mobile phone shells,and analyze the prominent effects of defects under the imaging light path of different light sources.Determine the optimal light source size and lighting method to make the three light paths correspond to the three RGB channels of the color camera.Then,the hardware such as camera,lens and light source is selected,and the imaging experimental platform is designed and built.The experimental results show that the imaging of speckle-like defects using coaxial light and the dark-field imaging of stripe light and scratches arranged in parallel is optimal.Compared with traditional methods,it can effectively save hardware space and cost,and is convenient for flexible configuration.(2)Design the image channel separation and image preprocessing method for the surface defects of mobile phone shells.The camera image was separated using RGB color space,and the types of image interference noise and the effects of various filtering algorithms were analyzed.The median filtering and Gaussian filtering were used to remove salt and pepper noise and Gaussian noise,and the histogram equalization method was used to enhance defect features.(3)Investigate the edge defect extraction and fusion defect detection algorithms for three-channel image surface defects.Analyze and compare the effects of Prewitt,Sobel,Roberts,LOG,and Canny algorithms on the edge extraction of mobile phone shell surface defect images,and the algorithm flow of three-channel image extraction feature fusion detection is designed.Combining the different characteristics of the defects to be inspected in the three imaging optical paths,a method based on mathematical morphology algorithms to design different structural elements for the defects to be inspected in each channel image,and then combining the results to achieve defect recognition.The image processing results show that the method works well,and provides an efficient detection algorithm for the single-camera to three-channel combined imaging method.Finally,aiming at the samples of mobile phone shell,taking three kinds of defects,such as horizontal scratch,vertical scratch and spot,as the detection objects,the defect fusion detection algorithm is realized based on MATLAB platform,and the detection results are compared and analyzed.The research results of this thesis provide solutions for other materials,colors and similar visual detection applications.