| Aluminum wire wedge bonder is an important production equipment of wire bonding process in packaging process.It integrates image recognition,automatic control and bonding technology to realize wire connection between semiconductor chip and lead frame.Because the position of the chip sticking to the pin frame is not fixed,there will be small position deviation and angle deflection,so the image recognition function is needed to accurately identify the position and tilt angle of the chip on the frame.The identification and positioning of the chip is one of the key technologies to realize automatic welding.In this paper,the gray-scale-based template matching method and feature-point-based matching method are studied,and a lot of image matching experiments are done.The main work of this paper is as follows:Firstly,this paper first introduces the aluminum wire wedge bonder.For the image acquisition part,the hardware system design is carried out according to the system requirements analysis.Secondly,due to the influence of noise,the real-time chip images collected by aluminum wire wedge bonder can not be directly used for chip matching.In this paper,the median filter is used to filter the template image and the image to be matched.According to the need,in the gray-based template matching method,this paper uses binarization method and Canny edge detection algorithm to preprocess the matched image,in order to make preparations for the Hough transform to detect the straight line.In the matching method based on feature points,this paper uses morphological filtering to preprocess the template image and the image to be matched.Morphological filtering can find the obvious maximum or minimum region,which is ready for the feature point detection method used later.Thirdly,the template matching method based on gray level is studied in this paper.Normalized square difference method,normalized correlation method and normalized correlation coefficient method are used to carry out matching experiments,and then the matching speed is improved by the pyramid hierarchical search method.Aiming at the problem that the gray-based template matching method can not recognize the rotation angle of the chip,this paper uses Hough transform to detect the straight line of the chip edge in the image to be matched,calculates the slope of the straight line to get the rotation angle of the chip,then rotates the template image of the chip to the corresponding angle,and then matches the chip through the experiment.This method is verified to improve the matching accuracy.Fourthly,for the chip with complex surface texture,this paper proposes a matching method based on feature points.The common SIFT,SURF and ORB algorithms are introduced in detail.A large number of experiments have been done to verify the feasibility of the matching method based on feature points,compared with the template matching method based on gray level.The matching method based on feature points can directly and correctly match the chip with rotation angle. |
