Algorithm for automated visual inspection of monolithic integrated circuits using neural networks

Download article in PDF format

Authors: Shiryaev B. V., Yuschenko A. Yu., Bezruk A. V.

Annotation: The article describes the problem of visual control in the production of MMIC. The algorithm for automated visual control is implemented using neural networks. The neural network performs a parallel analysis of micrograph and photo-mask device. Then it outputs information about the defectiveness of the device. Thus, the speed of visual control of MMIC increases without losing the quality of defect detection.

Keywords: automation, mmic, convolutional neural networks