Trademark represents the quality of the goods and the manufacturer's reputation, which is an important symbol of goods; automatic trademark recognition has a wide range of application prospects in many fields such as e-commerce, advertising and logistics transportation. In target recognition algorithm, the most critical is to extract the target feature. Because of the moment feature has better invariant characteristics; it is widely used in image feature extraction. In this paper we use Hu modified moments and the minimum distance classifier for trademark recognition, the experimental results show that the modified moments have higher recognition rate than the original Hu moments, this method can improve the recognition rate of the target image, it is conducive to trademark recognition.