基于神经网络的IC芯片图形缺陷检测技术研究

Integrated Circuit Chip Defect Detection Technical Research Based on Convolutional Neural Network

基于卷积神经网络的IC芯片图形缺陷检测方法,针对具有缺陷特征的图形图像样本集进行机器深度学习训练,可实现对IC芯片图形中如断线、起泡、腐蚀、划痕、裂纹、污染、崩边等图形缺陷的识别和区分。实验证明,这种方法可用于集成电路芯片的图形缺陷测试。

Based on the Convolutional Neural Network(CNN),the graphic defect detection method of IC chip can realize the identification and discrimination of graphic defects such as broken lines,blisters,corrosion,scratches,cracks,pollution and edge breakage in IC chip graphics by carrying out machine deep learning training for the graphic image sample sets with defect characteristics.The experiment results show that CNN can be applied to the graphic defect testing of IC chips.