In order to reduce the false detection rate of trademark retrieval algorithm, a trademark retrieval algorithm combining Zernike Moment(ZM) and Scale Invariant Feature Transform(SIFT) is proposed. The proposed algorithm consists of offline database construction and online retrieval. Firstly, Zernike Moment and SIFT features are extracted from the query image. Then, according to the similarity measure between the feature set of the query image and the Zernike moment feature set of the images stored in the database, the candidate trademark set is formed. Finally, the SIFT feature is utilized to detect the query image and candidate image accurately, and the similarity distances are sorted, and the result is returned to the user. The experimental results show that, compared with the current popular trademark retrieval, the proposed algorithm has excellent retrieval performance and is effective in scaling, translation, blurring, perspective, skew, distortion and other transformation forms. The scheme shows good Precision-Recall curve and F value, good real-time performance, strong robustness, and has a certain practical role in trademark registration, auditing and protection.