摘要：Aiming at pitting of gear failure forms and attrition, this paper uses image processing software of MATLAB and the research image extractive technique of image character. Based on the two different image expressions of defect forms, this paper abstract defect area ratio as well as the characteristic value and abstract contrast with gray co-occurrence matrix as image characteristics. In order to verify the abstracted image characteristic for evaluating pitting and wearing effectiveness, we use the neighboring method to construct a recognition model, and analyze the collected image samples. The result is correct, but the recognition accuracy needs to be improved. At last, eccentricity ratio and circularity is abstracted as the characteristic value based on image morphological characteristics.
关键词：gear defection;image processing;characteristic value
