The Research Of The Defect Detection Technology For Display Based On Visual Perception

The quality inspection of display products plays an important role in the production process.The color defect of the display makes the detection accuracy of the existing optical detection equipment low or difficult to detect.Therefore,thesis focuses on the research of display defect detection algorithms such as color multi-point defect,line defect and weak cross defect.A software system combines with display defect detection platform is developed and applied in related production lines.The main work is summarized as follows:(1)Aiming at the low accuracy of color multi-point defect detection under complex background,a color point defect detection method based on color channel transformation and feature fusion segmentation is proposed.The HSV color space is used to construct human visual perception image,and the constraint coefficient of RGB space channel is optimized.The local contrast method is used to extract color channel features.Finally,the adaptive threshold segmentation method of feature fusion is used to obtain the defects.The experimental results show that the defect detection accuracy of the proposed algorithm is 94.6% and the recall rate is 98.1%,which is significantly improved compared with the traditional detection methods.(2)Aiming at the problem of low contrast color line defect detection,a color line defect detection method based on significant color feature fusion was proposed.The maximum relative entropy criterion is used to obtain the color significant channel in the opposite color channel.Using color feature fusion is to obtain salient image,and the color line defect detection segmentation implementation.Experimental results show that the detection accuracy of the proposed method is more than 90%,which is obviously better than the performance of existing related algorithms.(3)Aiming at the problem that weak crosstalk defect can’t be detected,a cross defect detection method based on frequency domain filtering of salient color channel is proposed.The maximum relative entropy criterion is used to select the salient color channel in RGBY,and the color quaternion matrix is formed with the Lab color space.The hypercomplex Fourier transform is used to transform it into logarithmic spectrum of significance in frequency domain.Finally,combining with Gaussian threshold band-pass filtering and inverse hyper-complex Fourier transform,the low contrast defect of display image is separated from the background.Experimental results show that the proposed method can effectively detect cross color defects of weak contrast.(4)According to the actual production requirements of display,a software system combined with online defect detection hardware platform is developed,which can provide automatic detection results and statistical analysis of performance parameters.The evaluation of color Mura defect detection index is expanded.The system has been successfully tested on related production lines,and has strong generalization and practicability,which can effectively improve the product quality control ability and economic benefits of manufacturers.