排除模板迭代算法在数字减影血管造影中的应用

本文基于ＤＳＡ图象中由于选影剂的影响存在灰度差别的特点，提出了一种新的匹配算法－－排除板迭代算法，成功地实现了ＤＳＡ图象中的运动伪象和灰度伪象的校正。在我们的算法中利用排除模板将存在显著灰度差别的ＤＳＡ序列图像转换为可以 利用古典匹配算法的序列图象，而且在用排除模板进行匹配的过程中，逐次迭代，直至最佳匹配状态。整个算法实现了全自动，且匹配速度快，搜索范围大，精度可达１个象素点。实验结果表明，此方法

Recursive Exclusion Template Algorithm for Digital Substraction Angiography

In order to obtain high-quality digital substraction angiography (DSA) images, several artifacts caused by motion and gray-level variations should be corrected before the substraction operation. Due to the existence of the contrast medium, we are prevented from using classic matching algorithms, e. g. CC (correlation coeffient), IM(invariant moment), SAVD(sum of absolute values of the differences) etc. So, we suggest a new algorithm called "Recursive Exclusion Template Algorithm" by which the inference of the contrast medium can be successfully suppressed and classic matching algorithm can be suitably utilized. The main idea of the ET is to exclude all those pixels which have high gray-level difference between Contrast image and Mask image. These pixels are likely to be the ones containing the contrast medium. Because of the uncertainty character of those pixesl which are on the original images with artifacts, the ET can not be established successfully once, therefore, recursive algorithm is introduced. Our RET algorithm is fully automated; its computation expenditure is cheap and its matching extent is great ; it also offers a matching precision of 1 pixel. The results of our experiments indicate that RET algorithm can make great improvement on the quality of DSA images and will have good application prospect.