Aesthetic Defect Inspection Of A Polymeric Polarizer Via Saturation Imaging

Polarizer,as one of the most important components of thin film transistor liquid crystal display(TFT-LCD),its performance has an important impact on the quality of liquid crystal panel.The appearance defects of polarizer will reduce the display quality of the whole liquid crystal panel,and even cause the whole panel to be scrapped.Therefore,it is of great significance to study the visual detection technology of polarizer appearance defects.Aiming at the slight transparent dents which are difficult to detect,a polarizer aesthetic defect saturation imaging detection method based on machine vision is proposed in this paper.The main research contents are as follows:1.A saturation level-guided image enhancement method for extremely-slight transparent defects is studied in this paper.Fringe structured light saturation imaging method can effectively improve the contrast of defect imaging,but the contrast is very sensitive to saturation and imaging parameters needs to be carefully selected manually.Based on the mechanism of saturation imaging enhancement,a new definition of structured light imaging saturation with translation,scale and rotation invariance is proposed.The empirical model of the relationship between defect imaging contrast and image saturation is successfully established,and based on the defect contrast data obtained under five saturation levels,the best saturation is estimated by parameter optimization method.More than 300 defect samples have been detected by saturation imaging method,and the recognition rate is 100%.The experimental results show that the guidance method is effective and easy to use,solves the problem of manual selection of parameters,and provides a better imaging effect for defect detection.2.The design and optimization of saturation imaging detection system for polarizer aesthetic defects is studied.Obtaining high-quality and high-resolution defect images is the design requirement of the detection system,which can reduce the complexity of later image processing algorithm.All the requirements of defect imaging are analyzed in detail,and the imaging system is optimized according to the experimental hardware,defect size and detection accuracy requirements.The main modules include light source,imaging camera and lighting path.The structure of the imaging detection system is easy to configure and maintain,and the selection and optimization of system components can guide the design of other systems.3.A three-dimensional shape measurement method of dents based on saturation imaging and regression model is proposed.The optical imaging simulation results show that the gray level data of the saturated image of the defect is correlated with the morphological parameters.The dimensionality reduction is performed on the defect saturated image and the data set measured and marked by confocal microscope,and the support vector regression algorithm is used to establish the relationship between the saturated image and the defect morphological parameters.then the regression prediction model is applied to estimate the morphological parameters of new defect samples.The manifold of saturated image dimensionality reduction data can clearly show the changing properties of the three-dimensional shape of the defect,such as depth and width and so on.The experimental results show that the average relative error of defect depth estimation is 3.64%,the average relative error of defect width estimation is 1.96%,the error of measurement consistency is 7.39%,and the defect shape time is less than 0.01 s.The measuring accuracy and speed of three-dimensional shape of defects meet the technical requirements of accurate classification of indentation defects.