中文摘要：

由于噪声和成像条件的不同,准确得到两幅图像中提取的点特征之间的对应性是基于点特征的图像配准方法的一个难点。为此,文中提出了一种新的基于点特征的图像配准算法。该方法结合图像之间的拓扑关系和点特征之间的相似性,并利用变换模型限制求解点特征之间的对应性,然后根据对应的点特征求解变换模型参数,配准两幅图像。实验结果表明,所提出的方法对传感器相近的图像之间的配准是正确有效的。

英文摘要：

Due to noise and different imaging conditions, it is a very difficult problem for image registration methods based on point features to find the correspondence of the points extracted from the reference image and sensed image. A new image registration method based on point features is proposed in this paper. It gains the correspondence of the points according to the topologic consistence, the similarity of point features and limitation of the transformation model. Then it gets the parameters of the transformation model according to the corresponding points and registers the two images. Experiment results prove that the proposed method is accurate and effective.