基于SIFT的印鉴配准算法研究

SEAL REGISTRATION ALGORITHM BASED ON SIFT

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。

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.