Research On TFT-LCD Mura Defect Detection Algorithm Based On Image Processing

With the continuous development and perfection of computer science and technology,digital image processing technology has been more and more applied to the automatic detection of industrial production.Compared with the human eye detection,image processing based on machine detection has higher objective consistency,which is faster,more efficient and more stable,and it can be applied to many industries.In the field of production and application of display devices,traditional CRT display has large volume,electromagnetic radiation and other serious shortcomings,TFT-LCD with no radiation and light has attracted more and more attention and has been deep into all aspects of people's lives,it has been widely applied to the mobile phone,game machine,TV,PC and other industry products.In recent years,with the pursuit of portability of liquid crystal devices,the size of glass substrates in the main components of TFT-LCD has grown larger and thicker.Although LCD manufacturers have been improving the production skills of TFT-LCD,it is still difficult to avoid some visual defects.Common TFT-LCD visual defects include defects,line defects and surface defects.Among them,the Mura defect which belongs to surface defects is the most complex and the most difficult to detect.At present,the commonly used detection method in the industry is artificial vision detection.But this method can not fully guarantee the stability of human's subjective judgement factors,and it is easy to appear problems such as leak detection,misjudge and so on.It resulting in low efficiency of artificial detection,which can not guarantee the production requirements of TFT-LCD.Therefore,the research and exploration of the automatic detection algorithm for Mura defects has become a problem that must be solved on the road of TFT-LCD development.At the beginning of this paper,it introduced the development trend and structure of TFT-LCD,the causes and characteristics of Mura defects,and summarized the previous research achievements and prominent detection methods in the field of Mura defect detection.Then the characteristics of TFT-LCD defect images are analyzed,and various image processing methods are used to highlight the Mura region,and finally the Mura region is separated.First,a gray enhancement method is proposed for low contrast and poor visual effects: histogram equalization and gray scale stretching,to enhanced image visual effect,making it easier to split Mura defects.Secondly,background suppression method is proposed for image background texture interference: Butterworth low pass filter and background reconstruction based on singular value decomposition,to eliminate the unique bright and dark background of the TFT-LCD screen and reduce the interference to the detection of Mura defects.Finally,the resolution of segmentation is affected by uneven illumination: gray equalization and homomorphic filtering,which eliminate the uneven illumination and facilitate the final segmentation of Mura defects.At last,it gives the detection process of Mura defects,introduces the aspects to be noticed at the actual Mura defect detection of TFT-LCD,experimental data show that the proposed method has a good detection rate.