影像識別技術在晶粒取放機定位系統之應用

Development of an Image Recognition Algorithm for a Die Pick-and-Place Machine

丁鏞(Y.Ting) ； 張義鋒(Y.F. Chang) ； 陳一信(Y.H. Chen)

《中原學報》 26卷1期 (1998/02) Pp. 39-51

https://doi.org/10.6358/JCYU.199802.0039

摘要

本研究主要的目的是發展一套高速影像識別法則，偵測出晶圓盤（wafer）中晶粒（die）之取放形心(centroid）與擺向方位角度（orientation)，並且回授相關訊號至X－Y工作平台控制器上，作為影像定位控制之用，以使得取放臂上的吸嘴能準確地吸取晶粒。由於整個取放的速度極快，除了對影像識別系統要求高速的識別能力之外，亦要求極高之精確度，以免吸嘴在吸取晶粒時，由於未對準形心而在高速移動下甩落晶粒。而本影像識別之方法採用離散最小平方法逼近 discrete least-squares approximation ) 晶粒邊緣，用以求得晶粒形心與擺向方位角度。

關鍵字

影像識別 ； 影像切割 ； 離散最小平方法 ； 動態臨界值

並列摘要

The objective of this study is to develop an image recognition algorithm, which is able to measure centroid position and orientation of the die on the wafer. Such information signals are transmitted to the controller on the X-Y table for positioning control, so that the bonehead installed on the robot arm can
precisely absorb the die. Since the completion of the pick-and-place process is limited in a few tenths seconds and the positioning accuracy is demanding, hence, the image recognition algorithm is in need of high speed and high precision. The discrete least- squares approximation method is proposed to determine the edge of the die first, then to estimate its centroid after the image processing.

並列關鍵字

imarge recognition ； discrete least-squares approximation method ； die pick-and-place