Kong Gang, Dai Dao-Qing, Zou Lu-Min
School of Mathematics and Computational Science of Sun Yat-Sen University.
Zhongguo Yi Liao Qi Xie Za Zhi. 2008 Jul;32(4):261-4.
In order to remove the artifacts of peripheral digital subtraction angiography (DSA), an affine transformation-based automatic image registration algorithm is introduced here. The whole process is described as follows: First, rectangle feature templates are constructed with their centers of the extracted Harris corners in the mask, and motion vectors of the central feature points are estimated using template matching technology with the similarity measure of maximum histogram energy. And then the optimal parameters of the affine transformation are calculated with the matrix singular value decomposition (SVD) method. Finally, bilinear intensity interpolation is taken to the mask according to the specific affine transformation. More than 30 peripheral DSA registrations are performed with the presented algorithm, and as the result, moving artifacts of the images are removed with sub-pixel precision, and the time consumption is less enough to satisfy the clinical requirements. Experimental results show the efficiency and robustness of the algorithm.
为了去除外周数字减影血管造影（DSA）的伪影，本文介绍了一种基于仿射变换的自动图像配准算法。整个过程如下：首先，在掩码中利用提取的Harris角点的中心构建矩形特征模板，并使用具有最大直方图能量相似性度量的模板匹配技术估计中心特征点的运动向量。然后，采用矩阵奇异值分解（SVD）方法计算仿射变换的最优参数。最后，根据特定的仿射变换对掩码进行双线性强度插值。使用所提出的算法进行了30多次外周DSA配准，结果图像的运动伪影以亚像素精度被去除，并且时间消耗足够少以满足临床需求。实验结果表明了该算法的有效性和鲁棒性。
