Electric wire is the basic support for economic development and state construction, as well as production and living of ordinary people. Cable material is an important component, which affects the security and reliability of electric wire. Super-smooth polyethylene thin film is a typical kind of insulating material, possessing low relative density, high corrosion and water resistance. However, defect generated during the machining process is the key factor for partial discharge. How to detect and control the surface protrusion is vital for electric power security. The original image of the surface protrusion is obtained by planar array CCD, the feature information of which is gained through morphology image process method. Defects are extracted and labeled by eight neighborhoods connectivity theory. The height and volume of the polyethylene thin film are calculated through triangle principle. All the experimental results indicate that the detecting program is effective.