Defect extraction techniques are studied regarding the silicon wafer surface defect. We design a new filter based on multiple structuring elements and suggest an improved markerbased and region merging watershed. To begin with, the filter which generalized close-opening and open-closing filter based on the morphological filter with multiple structuring elements is introduced to eliminate the noise and simplify the image and morphological gradient image while preserving the details. And then in order to reduce the over-segmentation of the watershed algorithm, this paper suggests an improved marker-based and region merging method, region average gray value and edge strength criterion is used in merging operation and has a good effect on segmentation. Finally, the improved watershed algorithm is applied to the filtered gradient image to get the defect contours. The experiments show that this method can eliminate the noises and extract accurately location and closed region contours, which lays a good foundation for defect feature extraction and selection.

No abstract

No abstract