The Key Technology Research Of Visual Detection On Chip Dispensing System

The highly speed development of the modern manufacturing industry urge the intelligent manufacturing and machine vision industry, the machine vision products provides great help to the manufacturer by their sustained and stable working style. At present, the application of more chip dispensing system on chip dispensing is a production method which is widely used in the industrial production. It not only has paste function, also can enhance the thermal conductivity of the electronic products. For the need for the production of dispensing operation, good accuracy, stability and long time uninterrupted work is the advantages of machine vision products which could not be replaced.The paper compares and analyzes the existing calibration methods, expounds the traditional calibration methods and self calibration methods on their advantages and disadvantages. Then use the Zhang’s calibration method to calibration the plane. During the calibration process, the choice of calibration board is irregular 11?9calibration target. In five different angles to take pictures, then use the Zhang’s calibration method to solve the problem. At last, using the maximum likelihood estimation method for distortion correction in the allowable error range planar calibration.To the chip of dispensing system for visual detection technology, according to the actual needs of the proposed an improved algorithm of center location has been used for the chip center location. First, using the center of gravity method and Gauss surface legitimate on the image of center location algorithm to process the images. Then, using the root mean square value method of data processing to inadequacies of the existing algorithm. Compared with the center of gravity method and Gauss surface legitimate on the image of center location algorithm, the proposed algorithm is more precise and less error than 3% in the center of the chip. The coordinate of the center of the chip was obtained through the front of the handle, using the method of least squares to the coordinates of the center, the coordinates of the center is transformed for the 3-D coordinates, control the motion to complete the final dispensing operations.Under the existing experimental conditions, through to the real production environment simulation experiments show that the coordinates from the center location algorithm proposed by this paper is more accurate, error is relatively smaller, can meet the requirements of industrial application.