中文摘要：

利用灰度信息对可见光与红外图像进行匹配时，其效果受两类图像间灰度分布差异的影响。结合这两类图像的特征，提出了一种基于边缘图像和SURF（Speed—Up Robust Feature）特征的图像匹配方法。首先采用改进的三次B样条分别对两幅源图像进行边缘提取；然后利用SURF算法在边缘图像上进行特征点检测；再通过最近邻次近邻比值法对特征点进行粗匹配，最后利用对极几何约束的RANSAC算法剔除误匹配点对，从而实现图像的匹配。实验结果表明，在正确匹配率方面本文算法明显优于Canny边缘提取和SURF的匹配方法，具有一定的有效性。

英文摘要：

Due to prominent distributional variations of grayscale between visible and infrared images, traditional matching methods based on grayscale information show obvious deficiency on these two kinds of image matching. Combining with the characteristics of these two kinds of images, an image matching algorithm based on edge image and SURF features is proposed. Firstly, we respectively extract edge images from the original images by adopting improved cubic b-spline. And secondly, we extract the SURF features from the edges of both images, then the ratio of the closest neighbor and second closest neighbor is used in the features matching. Finally, the RANSAC algorithm is applied to remove false matching points. The experiment results show that the proposed method is better than the Canny and SURF method in the correct matching probability, and the validity of matching method proposed is proved.