中文摘要：

针对心脏周期剧烈运动产生的冠状动脉血管非刚性变化所引起的造影图像处理误差，提出一种基于快速独立分量分析（FastICA）的动态冠脉造影图像序列特征提取方法，改进了以往算法将动态序列中的每帧作为静态图像单独处理而未考虑帧间关系的缺点，综合序列中各帧信息进行分析，解混分离出背景区和感兴趣区。在传统FastICA算法基础上，提出一种适合冠脉造影图像灰度级分布特征的自适应初始权值计算方法，在保证效果一致的情况下，使迭代次数降低约45％。最后，设计一种基于感兴趣区域的伪彩色编码方法。

英文摘要：

Aiming at the angiographic image processing differences caused by cardiac non-rigid motion, a feature extraction algorithm of moving coronary artery angiograms based on fast independent component analysis is put forward. In the traditional schemes,every frame of the dynamic sequence is processed separately as a static image without considering the relation of them,which is improved in the novel algorithm. Comprehensively analysing every frame of the dynamic sequence, the background region and artery region are separated by the demixing technique. And an adaptive initial weight algorithm according to the gray distribution of the coronary artery angiogram is proposed. The iteration times is reduced 45 below the premise of the same result. At last,a pseudo-color coding method based on salient object detection is designed.