针对传统的红外与可见光图像融合算法提取目标信息不突出的问题, 提出一种基于非下采样剪切波变换和稀疏结构特征的融合方法.首先用非下采样剪切波变换分解源图像; 然后通过主成分分析提取低频子带系数中边缘和轮廓等显著特征, 引导低频成分融合规则的设计, 同时基于结构信息的稀疏性指导融合高频子带系数; 最后经过非下采样剪切波变换逆变换得到融合后的图像.实验结果表明, 该方法在保留可见光图像背景信息的基础上, 突显了红外图像的结构信息, 有效提高了融合效果.

Since the object information can not be extracted efficiently by the traditional infrared and visible image fusion algorithms, an infrared and visible image fusion method based on the non-subsampled shearlet transform and sparse structure features is proposed. Firstly, the source images are decomposed by the non-subsampled shearlet transform. Then, benefit from the advantage of principal component analysis on extracting edge and contour significant features, the fusion rule in low-frequency sub-bands coefficients are merged by using the principal component analysis-based approach. Afterwards, the sparseness based on structural information guides the fusion of high frequency subband coefficient. Finally, the inverse non-subsampled shearlet transform is employed to obtain the fused image. The experimental results demonstrate that the proposed method preserves the background information on visible image and highlights the structural information on infrared image, and improves fusion results effectively.