Research On The Vision System Of LED Automatic Die Bonder

As the application of LEDs continues to expand into new market areas,people begin to invest more energy in improving existing equipment and developing new equipment.LED automatic die bonder is the equipment used to realize the die-bonding link in the LED packaging stage.Die bonder sucks the chip through the soldering head and then transfers the chip to the pad of the LED bracket.The vision system of LED automatic die bonder determines the speed and accuracy of the equipment's picking and fixing chips,and mainly completes the identification and positioning of LED chips and chip defect detection function.This thesis focuses on the in-depth study of the vision system of LED automatic die bonder,builds the hardware platform of the system and an automatic selection system for the chip template of LED die bonder.And according to the characteristics of LED chip,we have studied two recognition and positioning algorithms,and put forward the chip identification and positioning algorithm flow of this thesis.Finally,different judgment criteria are constructed to detect various chip defect types.The research content of the thesis mainly includes:(1)The important components of the vision system of die bonder are selected,and the hardware experiment platform for the vision system of LED die bonder are built.The software requirements of the vision system of LED die bonder are analyzed,and the LED chip identification positioning and defect detection process are proposed.(2)An automatic selection system for chip templates of LED die bonder is built,which can automatically obtain qualified chip templates and accurately identify LED chip characteristic parameters,such as chip spacing,length and width,area etc.The chip template automatic selection system of the die bonder first uses the projection positioning algorithm proposed in this paper to project the collected multiple chip images,choose the chip image with the lowest projection standard deviation value as the best chip image,then the chip template was obtained by processing the best chip image.The specific method is to segment the image by threshold value and calculate the connected area,then exclude the maximum and minimum area,and select the chip which is closest to the average area as the chip template.Then,based on Freeman chain code,the chip contour is tracked to determine the four corners and boundaries of the template chip.The length,width and chip spacing of the chip template are obtained according to the corner coordinates.Finally,the chip template is changed by color inversion and the connected area is obtained to obtain the pattern characteristics on the chip.(3)Two LED chip recognition and positioning methods are studied,one is the chip identification and location algorithm based on regional features,the other is the chip identification and location algorithm based on gray information.The chip identification and positioning based on regional features are carried out according to the gray characteristics of the chip,and the threshold segmentation is carried out.Then the chip area and the connected area of each small point on the chip are compared with the characteristic parameters of the chip template to realize the qualified chip identification.Finally,the chip is located by Freeman chain code contour tracking.In the chip identification and location based on gray information,NCC algorithm is used to match the chip images first,and then the similarity measure values of all the matching boxes of each chip are sorted from the largest to the smallest,and the matching box with the largest similarity measure value is selected as the chip best matching box to complete the chip identification and location.And through experiments,finally choose the identification method based on gray information,design the chip positioning process.Finally,different judgment criteria are constructed to detect five types of chip defects.