This work aims at detecting defects on metallic industrial parts with complex surface. The searched defects are scar, rust and inclusion. A specific inspection system has been designed to deal with the particular inspected surface features. In the system, two images are acquired with the help of multiple light sources and CCD color digital camera. Based on the traditional algorithm, the background removal algorithm is designed in this article, and the color image feature extraction is also used for auxiliary analysis. A thresholding processing is then applied on this image in order to segment the imperfections. The perimeter and area of defects are calculated to further identifies the characteristics. The developed inspection system has been tested and it can accurately detect the defects of industrial parts with complex background. The recognition rate of algorithm is more than 96%.