Abstract
The infrared image of solar cell's electroluminescence (EL) is one of the important means of hidden defects detection. In order to improve the automatic recognition rate of defect images, this paper adopts improved invariant moments for feature extraction. The scale factor of the improved invariant moments is eliminated by transformation. Therefore they have the properties of translation, rotation and scale invariance simultaneously in discrete state. At the same time, Support Vector Machine (SVM) is used to distinguish the defect image. The system which combined invariant moments with SVM is applied to classify the debris, crack, off-grid, open weld and black pieces. The recognition rate of 5 kinds of defects has reached more than 90%.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have