Seal registration is the foundation and the key of seal identifying algorithm .Traditional seal registration algorithm based on tem-plate and feature is sensitive to noise , and is difficult to solve the problem of multi-scale matching.Therefore, we introduce the SIFT algo-rithm which has strong ability of multi-scale adaption and denoising property to seal registration , and then combine it with some denoising al-gorithms such as RANSAC to form a new seal registration algorithm .Experimental results show that the new algorithm greatly improves the matching accuracy and robustness .Besides, it has a better performance in multi-scale matching than the traditional algorithm .%印鉴配准是印鉴识别算法的基础和关键。由于传统的基于模板和基于特征的印鉴配准算法对噪声较敏感，同时难以解决多尺度匹配问题。因此，引入具有尺度不变性且抗噪性较强的SIFT （ Scale-Invariant feature transform ）算法进行印鉴配准，再结合RANSAC等去噪算法形成新的印鉴配准算法。实验结果表明，新算法极大地提高了配准的准确性和鲁棒性，并且解决了传统算法中的多尺度匹配问题。
展开▼
