Convolution neural network based mode selection and defect classification for image fusion

摘要

Disclosed are systems and methods for classifying defects using hot scan and convolutional neural networks. A primary scan mode is identified by the processor and a hot scan of the wafer 14 is performed. The defect of interest and nuisance data are selected, and images of those regions are captured using one or more secondary scan modes. An image set is collected and divided into subsets. The CNN is trained using the image subset. An ideal secondary scan mode is determined and a final hot scan is performed. Defects are filtered and classified according to the final hot scan and ideal secondary scan mode CNN. The disclosed system for classifying defects utilizes a processor and electronic database in addition to an image data acquisition subsystem such as a scanning electron microscope.