Abstract: Because carbon steel materials are widely used as the grounding grid materials of substations in our country nowadays,which are easily corroded and the error of human judgment for corrosion is large,this paper proposed a methodof using image processing technology to deal with the color and features extraction.Firstly,nonlinear restructuring of luminance and chrominance was used to enhance the corrosion image in the luminance/chrominance color space.Secondly, the main features of an image,including color,texture and gradient magnitude, were measured by using the local homogeneity,Gabor filter and color spaces.Then,artificial bee colony algorithm was used to get the best seeds and the best similar values of image,and seeded region growing was used to segment image into small areas of corrosion.Finally,the similarity of these small corroded areas was measured.Thereby the analysis of corrosion levels was completed.

Key words: Feature extraction,Artificial bee colony algorithm,Seeded region growing,Similarity measure