HISTOGRAM-BASED METHOD FOR AUTO SEGMENTATION OF INTEGRATED CIRCUIT STRUCTURES FROM SEM IMAGES

摘要

A histogram-based method for auto segmentation of integrated circuit structures is disclosed. The method includes an auto-segmentation process/algorithm, which works on the histogram of the SEM image and does not try to model the noise sources or the features. The auto-segmentation process/algorithm extracts the number of peaks in the histogram from low magnification SEM images or SEM images not necessarily having high quality images, significantly simplifies the traditionally lengthy and expensive IC reverse engineering efforts. Hence, the size of the image does not affect the final segmentation. The auto-segmentation process/algorithm performs the steps of: extract a first histogram from the first SEM image; identifying boundaries of the plurality of structural elements in the IC based at least in part on an output of the first histogram; and auto-segmenting the first SEM image into the plurality of structural elements.