Detection of cladding ice on transmission line based on SVM and mathematical morphology

In this paper, we propose a method based on support vector machine (SVM) and mathematical morphology to detect cladding ice on the transmission line in surveillance image. Firstly, we construct a sample set for SVM training process and obtain the classification model. Secondly, the original image pixels are classified into two categories by using of the SVM classification model, based on which we get the initial segmentation image. At last, mathematical morphology operation is used to denoise the initial segmentation image and then the final segmentation image is obtained. The experimental results show that our method can precisely detect the ice region in the original image.