Research On Detection Technique For Mura Defect Of TFT-LCD

Thin film transistor liquid crystal display(TFT-LCD)in the technology and production of rapid development,is gradually to light,large-size,high-resolution development.In the large size and large-scale production,the probability of LCD panel appears a variety of defects is also greatly increased at the same time,which not only limits the production,but also increases the cost.Aiming at the problem that the Mura defect on the TFT-LCD panel is low-contrast,the blurred edge and the irregular shape.After attempting a large number of detection methods,the paper focuses on studying and designing a TFT-LCD Mura defect detection technique under the uneven background brightness.According to the actual detection needs,the detection process includes Gaussian filter denoising,bicubic B-spline,Otsu-based piecewise exponential transformation with double ? to enhance contrast,Otsu's method threshold defects segmentation and SEMU standard quantitative evaluation.The key technology of each stage of the detection process is analyzed and deduced,and compiled with Matlab to achieve.The following points are mainly studied:(1)First,the definition,causes and classification of Mura is summed up,and then the source of the image noise and four commonly-used denoising methods are summed up.(2)In the part of surface fitting to remove the background,the theoretical knowledge of the binary cubic polynomials surface fitting and the bicubic B-spline surface fitting are summarized.The multiplicative equation is used to calculate the bicubic B-spline surface and the fitting data are compressed to improve the speed and computational efficiency of the algorithm.In order to improve the accuracy of the traditional bicubic B-spline fitting,the smoothing of the fitting surface is carried out to reduce the influence of Mura region point to interfer local shape of the fitting surface.(3)Two kinds of contrast enhancement methods are introduced and their shortcomings are analyzed,then a new enhancement method is proposed: automatically select the threshold as the segmentation point of the piecewise function by introducing the largest interclass variance(Otsu)method.The ?-index transform is introduced to make different gamma transforms for the background and the target region.The proposed method can enhance the target area contrast and suppress the changes of gray value in the background area,while enhancing the edge of the Mura defect.And thenOtsu's threshold segmentation is used to binarize the image,and the defect area and other related parameters are extracted.(4)The SEMU standard is introduced,and the relevant parameters extracted in the previous stages are brought into the formula to obtain the Semu value for the rating to determine the accuracy of the algorithm.Combining contrast enhancement and defect segmentation to show the effect of each algorithm,it proves the practicability and efficiency of the proposed algorithm.Through the Mura defects detection experiment,summed up the detection algorithm flow.The experimental results show that the proposed algorithm can effectively detect all kinds of common Mura defects.For 100 samples,the effective detection rate is 99% under the relevant parameters.