Application Research On Machine Vision In RFID Tag Package Equipment

As ICs continue to become more sophisticated and minimum feature sizes keep shrinking, the need for higher accuracy in fine pitch interconnect applications has become more important then ever. Machine vision can identify and locate the micro chips through optical system design and image processing technology, so it is widely used in semiconductor packaging equipment. For smaller size, higher accuracy and higher speed for RFID tag package equipment, the hardware implement and image processing of vision alignment system are more complex than ever. The research work of my paper is based on RFID tag package equipments which is developed independently by our group. The implement of on-the-fly vision alignment system and corresponding image processing technology of equipments are studied systematically. The main contents of my paper are as follows:Multi image processing methods are studied to enhance the adaptability and flexibility of vision system. Firstly, smoothing-filtering approaches to the fiducial mark are introduced in my paper contain median filtering. Secondly, threshold segmentation approaches to the image contain simple iterating method of optimum threshold segmentation and maximum-minimum variances segmentation methods. An auto threshold algorithm is designed to separate the feature from the background. Finally two methods of normalized grayscale correlation are compared. For failing to locate the image for missing features of the fiducial mark, an efficient approach to the image locating is designed for it.A multiple reflector vision system scheme is designed to meet the requirements of high speed mounting of the components. Calculations for bias and rotor center of bonder header are discussed in details. This part also analyzes speed and accuracy of the system. Experimental results demonstrate that they can meet the need of higher accuracy and real-time performance.Based on the working principle of the equipment, the paper discusses the working flow of three sets of vision systems. The hardware construction and software development of multi vision systems are introduced in detail. Considering allocation of system resources, the multi thread technology is proposed in vision development.The content of the paper, can be as the research information of the machine vision's application in the packaging of the semiconductor, and can be the reference as the part of machine vision in other robotized productions.