APPLYING LOCAL REGION DESCRIPTORS AND SOM CLUSTERING ALGORITHM IN LED DIE COUNTING

Tang Liang,Liu Ye,Chen Yanqiu

DOI: https://doi.org/10.3969/j.issn.1000-386X.2012.10.006

2012-01-01

Abstract:Traditional image processing algorithms tend to be affected by light interference,angle variation,noise and other factors,making it impossible to get identification results accurately and quickly.This paper presents a novel local region descriptor to describe the gray value feature of the target point group.When combined with self organising map(SOM) clustering,it is able to identify the die area in light-emitting diode(LED) efficiently and to calculate the die number accurately.Experimental results show that this algorithm is capable of counting the LED die number efficiently with higher accuracy and better robustness.