基于高曲率特征点匹配的红外可见光图像配准

Infrared-Visual Image Registration Based on High-Curvature Feature Matching

红外与可见光图像的融合扩展了视场信息的频谱波段,弥补了传统视觉成像的不足。根据红外图像的成像特点,研究了红外和可见光图像的配准方法,提出了一种基于高曲率特征点匹配的图像配准算法。首先,对红外和可见光图像进行边缘检测,在边缘上提取曲率变化的局部极值点;然后,根据特征点的空间相对关系,利用改进的相似三角形匹配法进行特征匹配;最后,实现了红外与可见光图像的配准和融合成像。实际场景试验表明,该算法可应用于实际工程。

The fusion between the infrared image and the visual image can extend the spectrum band,and make up for the deficiency of the traditional visual imaging.According to the imaging characteristics of infrared images,image registration methods between the infrared image and the visual image are researched.An image registration algorithm based on high-curvature feature matching is proposed.Firstly,edges of the infrared image and the visual image are detected.The local extremum dots are picked up by curvature change.Secondly,according to the space position relations between the featurs,the festures are matched with an advanced method of similar triangle matching.Finally,the image registration between the infrared image and the visual image,and the fused imaging are realized.Experiments of real scenes show that the algorithm can be applied in practice engineering.