The Research On ACF Particles' Automatic Optical Inspection Technology In COG

With the rapid development of electronic products, liquid crystal device is widely used. The production processes for liquid crystal device is some complex in which defects may appear and there's much difficulty in automatic detection, so it is usually manually detected for the defects of liquid crystal device. As with advancements in science and technology, there is growing demand for automated detection, manual detection cannot meet the market requirements, so in recent years the automation of industrial production has been research focus in various industries, which is also the inevitability for industry transformation.On the market, a technology named Chip on Glass(COG) is used in most connections between the IC device and liquid crystal cell glass, COG uses an Anisotropic Conductive Film(ACF) to connect the IC device and liquid crystal cell glass. Current detection method is mainly artificial viewed with a microscope by checking the ACF conductive particle number on liquid crystal cell glass pins which is of very low efficiency and of great error probability. In this dissertation, the automatic detection method of ACF particles can solve the problem of artificial detection, the detection efficiency and accuracy can meet the detection standard of industrial field.In this dissertation, the author firstly makes a brief introduction to the use of ACF in COG, and a detection method based on machine vision is proposed. In machine vision detection, the method of digital image processing is mainly used, which is safe and reliable.Secondly, the image of ACF particles is analyzed. Particle image contains local maximum and local minimum of gray value, detection of particles according to the change of the maximum and minimum values of gray value.Finally, analyzing the relationship between the number of particles and the area of the particle, and the conclusion is drawn that the number of particles is proportional to the area of the particle. Based on this, a detection method based on the area of particles is proposed. Because the positions of the maximum and minimum value of ACF particle's local gray level are relatively fixed, according to the direction of light, a differential method has been used to enhance the image in this dissertation, which makes the image feature more obvious, and brings great convenience to the detection. Detection method based on particle areas does not rely on a specific pixel gray value, so it has wide applications, especially when the image is very seriously noised, still has a very good detection effect.What's special about this dissertation is that it proposes two new detection algorithms for particle image, which are more efficient than the traditional template matching method. By accelerating the algorithms, the efficiency of the algorithms is about 20 times higher than that of the template matching method, and it can fully meet the requirements of the automatic detection in industry. To our knowledge, it is the first report that using these two methods to detect ACF particles in china.