K-means clustering with morphological filtering for silicon wafer grain defect detection

Grain surface defects have become an important factor affecting chip quality. How to correctly detect grain surface defects is a complicated and challenging task. Defect detection technology has become a key technology in the chip industry. K-means clustering algorithm is used for image segmentation. However, the traditional K-means clustering algorithm is sensitive to noise. In order to improve the robustness of K-means clustering algorithm to image segmentation, local spatial information is often introduced into the algorithm. This paper proposes an improved k-means clustering algorithm based on morphological filtering. By introducing morphological close operation and open operation into the k-means clustering algorithm, image noise immunity and detail preservation are guaranteed. Experiments show that the improved algorithm has good segmentation effect and is more robust to noise than the traditional k-means clustering algorithm. The improved algorithm can effectively detect grain surface defects with accuracy of 99.02%.