Research On Automatic Probe Station Location Technology Based On Machine Vision

In recent years,with the rapid development of the semiconductor industry manufacturing process,The market demand of tiny chips has been increasing dramatically,the chip size,pad diameter and the distance between chip leads have been decreasing,resulting in higher demand for the motion precision of packaging equipment.The traditional packaging equipment has been unable to meet the demands,so machine vision technology has gradually become the key technology of IC(Integrated Circuit)industry.This paper mainly studies the development of machine vision and its application in wafer inspection.Based on image processing algorithm,the specific scheme of positioning system is designed.The paper includes the following contents:In chapter 1,the paper describes the application and research status of machine vision system,analyzes the application of machine vision in wafer inspection,and gives the framework of the paper.In chapter 2,according to the main functional requirements automatic positioning system,the vision positioning system of automatic probe station is designed.According to the requirements of system precision,the type of camera and lens in the vision system are determined.In chapter 3,we introduce four classical edge extraction algorithms in detail,and propose a multi-scale edge detection algorithm.Based on the multi-scale structure edge detection and the least squares technique,the wafer center detection scheme is implemented.The feasibility of the scheme is verified by experiments,and the repeated positioning accuracy of the center is 44?m.In chapter 4,the SURF and SIFT algorithms are studied based on feature point matching,and the advantages and disadvantages of the two algorithms are illustrated by comparing the two algorithms.At the same time,it is verified that the SIFT algorithm is invariant to rotation scaling.According to the experimental results,the SIFT algorithm is used to calibrate the wafer,and a specific alignment scheme is designed.The accuracy of the scheme is verified by experiments.In chapter 5,morphological image processing and several contour center algorithms are studied.Based on the morphological image processing algorithm and central moment algorithm,the detection scheme of the center positions of needlepoint as well as needle marks is determined,implementing the automatic needle-on-needle function of this system.In chapter 6,this chapter summarizes the main contents of this paper,and prospects the follow-up research work.