Abstract: The traditional fusion algorithms of transform domain image can’t clearly describe the edge of contour and the directional texture, and the linear nonstationary signal is usually unable to obtain a good result. In view of the above problems, this paper presents an image fusion algorithm that combines Contourlet transform and image regional features. This algorithm, based on the feature of high-frequency region and low-frequency region in Contourlet space, applies the method of energy feature matching in low-frequency region and fractal feature matching in high-frequency region to fuse. The experiments show that this algorithm performance is better than the traditional Contourlet fusion and wavelet fusion.

Key words: Contourlet transform, regional?features, image fusion

摘要： 针对传统变换域图像融合算法存在边缘轮廓和方向性纹理描述不够清晰，以及对线性非平稳的信号通常不能取得很好效果的问题。提出一种将Contourlet变换和图像区域特征相结合的图像融合算法。根据Contourlet变换空间下图像高频区域和低频区域的特点，采用低频区域能量特征匹配和高频区域分形特征匹配的方法来进行融合匹配。实验表明，该算法比传统的Contourlet融合和小波融合效果表现更好。

关键词: Contourlet变换, 区域特征, 图像融合