中文摘要：

红外光与可见光处于不同波段，其图像间的相关性较小。传统的基于特征的图像配准方法（如利用角点、边缘点等），在特征点选择时容易造成误匹配，这是由于有时特征点问的距离比较近造成的。针对此问题，本文提出了一种基于图像轮廓特征的红外与可见光图像配准方法。首先通过设置目标过滤器来提取明显的轮廓，再利用部分Hausdorff距离对轮廓进行匹配，计算出匹配轮廓对的面积和质心，并以此作为配准依据来对两种不同的图像进行配准。然后通过实验证明该方法的配准精度更高且克服了特征点误匹配的难点，这就可以解决刚性变换中红外与可见光图像间的配准问题。

英文摘要：

Because infrared light and visible light are in different bands, there are less correlation ：between an infrared image and a visible image. Traditional image registration methods based on features such as corner points and edge points etc. may result iu mismatching when the feature points are chosen. According to this problem, an infrared and visibie image registration method based on contour-features is proposed. Firstly, obvious contours are extracted from the images by using a target filter and are matched by using the Huasdorff distance. Then, the area and centroid of each pair of matched contours are calculated and are used as the registration rule for matching two different images. Experimental result shows that this method has a higher registration accuracy and has overcome the difficult that the feature points are easy to be mismatched. Thus, the registration of infrared and visible images can be implemented in rigid transformation.