Research On Visual Detection Algorithm For Mura Defect Of AMOLED Display Panel

Active matrix organic light-emitting diode(AMOLED)has become the symbol of high-end display technology because of its excellent performance in power consumption,contrast,response speed and so on.However,the ubiquity of Mura defects limits the yield of AMOLED display panels.The traditional Mura defect detection relying on the human eye is not only costly,but also affected by factors such as the fatigue of the tester.Although the defect detection of display panel based on visual algorithm has attracted the attention and research of many experts and scholars,the current algorithms generally have the problems of low accuracy,slow speed and poor applicability.In order to solve these shortcomings,this dissertation designs the Mura defect detection system of AMOLED display panel,and studies various algorithms for defect detection.According to the content and design requirements of the project,the Mura defect detection hardware system of AMOLED display panel has been built.In view of the high requirements for image quality in post-processing,this dissertation investigates the selection of hardware devices such as industrial cameras,industrial lenses,module generators and so on.Furthermore,the moiré pattern formed by the beat between the CCD sensor of the camera and the display screen is studied,and the corresponding hardware scheme is designed to suppress the moiré pattern.The preprocessing of the display panel image is very important for the defect segmentation detection,so this paper has carried out research and design on this.The unevenness of the collected images were corrected in the case of eliminating dust and noise interference.Aiming at the tilt problem of display screen image caused by the error of the image acquisition system,the tilt correction and extraction algorithm of target area of the image are studied and designed.In addition,this dissertation has also studied the background texture of the image caused by the physical structure of the display screen,and used relevant algorithms to suppress the background texture.Focusing on the detection difficulties of Mura defects,the segmentation detection algorithm and quantitative evaluation method for defects are designed.By analyzing the principles of various algorithms of defect detection,a neighborhood differential filter threshold segmentation algorithm is used to segment point,line and group Mura.Furthermore,a cascaded algorithm based on gray level difference of sub-image and region scalable fitting model is proposed to segment Mura defects,and is verified by simulation.In addition,this dissertation also uses the internationally accepted defect quantification standard to realize the parameter extraction and quantification of defects,and makes a graded assessment.The research results of this dissertation have a certain guiding significance for improving the yield of AMOLED display panel,and are helpful for the research of the causes and detection methods of various defects.