基于曲波变换的红外与可见光图像融合算法

Infrared and Visible Images Fusion Algorithm Based on Curvelet Transform

针对小波不能有效捕捉图像轮廓的不足,提出一种基于第2代曲波变换的图像融合算法。近似分量计算采用加权平均融合规则,细节分量计算采用像素级多分辨率融合扩展框架和对比敏感带通函数融合规则。实验结果表明,该算法在保留源图像边缘轮廓、抑制噪声方面均优于小波,融合图像更符合人眼视觉特性。

Aiming at the shortcoming of wavelet which can＇t efficiently capture image contour, this paper proposes an image fusion algorithm based on the second generation curvelet transform. It proposes a fusion rule containing weighted mean approximate coefficients and details coefficients combined with extended pixel-level multiresolution fusion framework, as well as contrast sensitivity band-pass function. It employs the subjective and objective evaluation to assess fusion results. Experiments indicate that the proposed approach is superior to wavelet in preserving source image edge contour and noise suppression, and fusion images better correspond with human visual perception characteristics.