In view of the current situation of transformer bushing monitoring of composite insulating material, the water drop image characteristic of transformer bushing is utilized, and the identification method of water repellent grade of Transformer Bushing Based on image feature is proposed.First of all, the methods of image denoising and en-hancement of wavelet theory, enhanced visual effects and details. Then the improved watershed method is used to segment the image, which ensures the authenticity of the original image, and the edge of the water is obtained. Ex-traction and selection of the most characteristic features. Finally, the least square support vector machine classifier is designed to realize the intelligent recognition of the water repellent level. The simulation results are given, and the method is applied to the on-line monitoring system, and it has achieved good results.