Study On Polarizer Defects Detection Technology Based On Machine Vision

Polymeric polarizer can easily form transparent aesthetic defects during production and transportation.As one of the important components of liquid crystal displays,their aesthetic defects have a serious impact on the display quality.For the detection of such defects,human eye recognition is still widely used.It is a great significance for the development of related industries to study the speed and the accuracy of machine vision detection technology.The transparent aesthetic defect of the polarizer is low in contrast and difficult to image under uniform light illumination.In this paper,binary stripe structured light illumination was used to enhance the defect imaging contrast.At the same time,the enhancement mechanism of defect imaging was simulated.Based on the above study,this paper further carried out the following research work:1.Inspection of extremely slight aesthetic defects.Experiments have found that such defects are still low in contrast even with streak structured light illumination.The simulation results show that the defects are extremely slight and difficult to image when the refractive index difference of the spot-shaped transparent defects is less than 0.0006 and the depth is less than 6 ?m.However,the imaging contrast can be significantly improved when the defects are located at the edge of the structured light.Then,this paper has its further research on the corresponding image processing algorithm--RPCA.And it's successfully to separate the defects from the black and white striped background.2.An aesthetic defects detection method based on structural light saturation imaging.Structural light saturation imaging can greatly improve defects imaging contrast,simplifying subsequent image processing algorithms.The active light two-step scanning was used to cover the entire polarizer.And the influence of the stripe width on the number of detection steps was experimentally analyzed.The corresponding three image processing algorithms are given in the paper.After comparison and analysis,it is found that the two-step average algorithm has greater advantages in speed and accuracy.3.A nondestructive method for measuring the depth of internal spot-shaped defects is proposed.The spot-shaped defects are modeled as micro-lens.Under the condition of fixed image distance,the correlation between defect brightness and defect depth during saturation imaging is experimentally studied.By using laser confocal microscopy calibration,an analytical expression for the depth measurement of defects was established.The measurement errors for the three samples were 1.2%,3.7% and 22.9% respectively.Preliminary experimental results show that this method can be used to estimate the depth of internal defects,so as to more accurately determine the quality of polarizers.4.Considering the size of polarizers and the cost of detection,a low-resolution imaging(120?m/pixel)inspection prototype system was designed and developed.For a 24-inch sample,four 500W-pixel cameras were used for full-coverage detection.Meanwhile,the image processing algorithm under low-resolution imaging condition is studied.In this paper,the RPCA algorithm is used to detect the extremely slight defects on the polarizer samples.And the defects are detected by the structured light under saturated imaging.The detection accuracy is high.A transparent defect depth measurement method based on correlation analysis is proposed for the first time.And a machine vision inspection platform for the defects of the polarizer is also built.The experimental results show that the detection method studied in this paper has a good application value and it can be used in the online defect detection of polarizers.