In semiconductor processing industry

manual calculation of number of wafer dies is inefficient and labour-intensive. In order to solve this problem

a machine vision-based autocounting method for wafer dies is presented. Firstly

the method uses Sobel operator to detect the edge of the wafer image. Then

row scanning and random sample consensus（ RANSAC） algorithm are used to locate the wafer. And then

improved RANSAC algorithm is used to roughly fit the dicing lines. Finally

the dicing lines are accurately fitted and the number of wafer dies is calculated by using the intersection points information of the dicing lines. The experimental results show that the method is stable and it has good efficiency and precision. It has good practical value for wafer dies auto counting.