中文摘要：

用偏振光显微镜旋转成像测量C/C复合材料热解炭消光角过程中,所采集的序列图像间经常出现目标偏移现象。针对该问题,提出了一种基于结构特征的图像配准算法。该算法采用圆边缘检测算子检测每幅图像中同一纤维圆截面的圆心,以圆心作为图像配准的匹配点,通过坐标变换实现图像配准。在热解炭消光角测量过程中的应用结果表明,该配准算法可以避免因目标偏移而造成的数据采集误差,保证消光角测量的准确实现。

英文摘要：

The object translation often occurs between the serial images, which are captured at different analyzer positions in the polarized light microscopy for determining the extinction angle （Ae） of pyrocarbon in C/C composites. The translation results in the incorrect Ae value. An image registration algorithm based on structure features was proposed for translation correction. A circular template was used in the algorithm to detect the centre of the same fiber circular section in each image as a matching point. The image registration was realized through coordinate transform. The proposed algorithm has been applied to the Ae measurement of pyrocarbon. The experiment result shows that the image registration can avoid the wrong sampling data caused by object translation and assure Ae measurement accuracy.