血管数字减影图象中的结构特征提取

Extraction of Structure Features in DSA Image

基于Ｘ线数字减影图象的三维血管结构重建很大程度上依赖于图象中血管的有效检测和结构关系描述。本文给出一个在局部特征检测基础上的结构特征提取和组织方法，涉及结构分析之前的中间视觉处理层次，即将图象中主要结构分离出来供结构分析之用。首先用一个突出结构检测算法独立地组织和选择中轴曲线片断和边缘曲线片断，然后将两者信号相结合以验证突出结构算法获得的结果并滤除噪声结构。

The overall scheme aiming at the three-dimensional reconstruction of vascular structures from X-ray projections relies greatly on the efficient detection and structural description of the vessels in the image frames. In this paper an approach for primitive organization and selection is proposed, which deals with an intermediate visual processing level, separating the principal structures from the image for structural analysis by using a structural saliency algorithm (SSA)to organize and select the salient centerline curves and salient contour curves separately and then combining them to verify the result of the SSA and to eliminate noisy structures.