A robust intelligence algorithm based on optimized K-Means for GPP contour extraction

for electronic components that are not packaged, traditional contour extraction algorithms do not work well because of the complex chip surface. This paper proposes a robust intelligent algorithm for GPP defect contour extraction based on optimized k-means. Firstly, the image is binarized by the optimized k-means algorithm based on the GPP gray value division of grain surface. Then, morphological operation and median filtering are applied to filter out the noise. Finally, canny operator's convolution template is used to calculate the amplitude gradient of the image to determine the boundary of the contour. Compared with Canny algorithm and Soble algorithm, the proposed algorithm has good robustness for GPP defect contour extraction and high boundary strength, which can be applied in industrial production.