Optimization of Wafer Scan Image Based on Improved Canny Edge Detection

The scanning electron microscope (SEM) is vital in wafer processing, providing high-res surface images for defect analysis. Despite optimizations, images may have noise and edge jitter from electromagnetic interference and mechanical vibrations. This paper aims to design a software-based image processing algorithm based on the Canny edge detection algorithm to address these issues as much as possible and enhance the readability of wafer images without compromising their original information. This paper mainly optimizes the algorithm from three aspects: The non-maximum suppression step is replaced by GEF to keep the edge information intact, the noise is suppressed by a special anisotropic Gaussian filter, and the threshold adaptive capability is added to the algorithm. The wafer image processed by the improved Canny algorithm will have less noise interference and more image details, which can reduce the workload of subsequent processing.