The Recognition And Classification Of The Hydrophobicity Of Composite Insulators Bead Image

Hydrophobic measurement is a main way to evaluate the performance of composite insulators. Detecting the surface hydrophobicity of composite insulators by digital image processing technology, we overcame the subjectivity of traditional methods. However, the complexity in accessing the hydrophobic image and the particularity of the hydrophobic image, which make the recognition performance of the hydrophobic bead image not ideal and affect the classification effect of insulator level. We researched the bead image recognition and the classification of hydrophobic level.In this paper, we discussed two kinds of bead image recognition algorithms: One is automatic threshold recognition algorithm of bead image by entropy, the other is bead image recognition algorithm based on Schr?dinger transform of image. The automatic threshold recognition algorithm of bead image by entropy, firstly, making the grayscale image of hydrophobic bead image into blocks, then competing bead image segmentation for each sub-graph by taking the method of the optimal entropy threshold and tracking the edge of bead, finally, merging the edge information of bead and recognizing the bead image. The bead image recognition algorithm based on Schr?dinger transform of image, which calculates the interior and exterior of the beads by using the Schr?dinger transform of the original image and its inverse image, by finding pairs of exterior and interior points with the smallest distance between them, contours of the beads are extracted.In terms of classified methods on the hydrophobicity of insulators, the common ways to identify the classification of hydrophobic level, whose basis of judging is single and accuracy of judging is not high, such as the shape factor method and the maximum water trace area ratio method. After completing the hydrophobicity of composite insulators bead image recognition, this paper also researches how to extract the features of single bead and bead image, then selecting features of bead image according to many factors, including the correlation coefficient between the hydrophobic level and the value of feature and the maximun, minimum, mean and variance of each feature in CH1~CH7 level, we use SVM support vector machine learning algorithm, training model by the selected features, establishing classification model, realizing automatic classification of the hydrophobic level of composite insulators.