Key Technologies Research Of AOI For Display Defect Inspection Of LCD

LCD has been widely used in the electronic information industry. Quality testingis an important part to ensure the quality of LCD products. The traditional manualinspection way has disadvantage of subjective and low reliability, which greatlylimiting the development of LCD industry. Automated optical inspection equipmentbased on digital image processing has the advantages of high precision, rapid speed,non-contact, which could overcome the drawbacks of manual inspection, and there willbe a good prospect in the liquid crystal display defect detection industry.To solve problems mentioned above, key technologies in automatic opticalinspection equipment of LCD display infects such as optical system design, imageprocessing algorithms and software system had been explored and researched in thisdissertation. The main contents are as follows:First, Based on analysising of recent research at home and abroad of the LCDdefects with automated optical inspection and researching of the principle of LCDdisplay and defect information and characteristics, the composition describedfunctional modules of the detection system had been elaborated. Focusing on analysisof the optical imaging system design, including the selection of the light source,camera, lens and confirmation of the parameters.Secondly, the image processing algorithm of defect detection had been researched.In the template creation stage, compared to the conventional method, the referencetemplate was generated by the average image of multi-view image after registration,and the tolerance template was generated by the standard deviation with expansion,which could greatest supprssing edge noise and increasing tolerance range of the edge;In image registration stage, according to the LCD imaging characteristics, registrationbased on shape matching was proposed. Edge contour feature extraction, featurematching, the search match strategy based on pyramid, image affine transformationhad been elaborate. Fast and accurate matching of the image was achieved; In the stageof results determine and information statistics of defect, algorithm based on dynamicthreshold segmentation and median filtering denoising had been used to defect decision, and connected region analysis had been used to important defect informationstatistical; The experiment results show that, the algorithm of defect detection hasadvantage of high precision, fast speed and stable detection.Finally, the software system for defect detection had been designed, whichconsists of four functional modules such as image acquisition, image processingalgorithms, standard library management, graphical user interface. The operation andfunction of each module was described, and the actual application was showed.This system had been tested in conventional LCD display defect detection, Theexperimental results show that the detection accuracy is0.1mm, and the accuracy rateis98%, it indicated that equipment practical value in this field. As far as we know, thealgorithm of shape-based matching registration and search match strategy based onpyramid used in inspection of LCD diplay infect have not been reported at home andabroad.