一种改进的SIFT算法及其在医学图像配准中的应用

An Improved SIFT Algorithm and Its Application in Medical Image Registration

图像特征点的提取是医学图像配准的基础,其精确性直接影响匹配的结果.目前在实际应用中常使用手工提取特征点的方法,精确性差且工作量大.SIFT算子具有良好的尺度、旋转、光照等不变特性,被广泛应用于图像配准中.由于SIFT匹配算法对特征点匹配的条件较为严格,特征点的数量常常无法满足医学图像配准的实际需要,并且存在一定的误匹配.为增加特征点的数量,提高匹配准确率,采用SIFT算法自动提取特征点,并使用特征点之间的Euclid距离作为相似性判定度量,根据医学图像的特点保留低对比度点,以实现医学图像的配准.实验结果表明该方法是有效的.

Feature extraction is the basis of medical image registration.The accuracy of feature points directly affects matching result.It often use manpower to do feature extraction at present,but the accuracy is poor and the workload is heavy.The features extracted with SIFT are invariant to image scale and rotation,and provide robust matching.SIFT algorithm has been widely used in image registration.As the SIFT algorithm is very strict about matching condition,the number of feature points often can not meet the needs of medical image registration,and there are some false matches to a certain extent.In order to increase the number of feature points and improve the matching accuracy,the Euclid distance has been adopted to determine the similarity between feature points and low contrast points have been retained according to the characteristics of medical images.The experiment results demonstrate the effectiveness of this method.