Segmentation algorithm of hydrophobic image based on region similarity
基于区域相似特征的憎水性图像分割算法

Hydrophobic measurement is an important way to evaluate the performance of insulators. Traditional segmentation algorithms may bring error segmentation in weak boundary, which affect the accuracy of confirming the level of hydrophobic. A new segmentation algorithm based on region similarity was proposed in this paper. According to the characteristics of hydrophobic image, gray similarity and gradient similarity were used in region growing. The final image segmentations were effectively obtained by mathematical morphology. The proposed approach achieves improved segmentation performance and provides evidences for contamination grades recognition of insulators.