基于Curvelet变换的图像融合改进算法

摘要

针对红外与可见光图像的融合，提出了基于二代Curvelet变换的图像融合改进算法。首先对两幅源图像进行Curvelet变换，得到其在不同尺度和方向下的变换系数。对于低频系数，根据红外与可见光图像的不同成像特点，采用基于局部统计特性的自适应融合策略；对于不同尺度和方向下的高频系数，采用基于局部区域能量匹配的系数选择方案。最后进行Curvelet逆变换得到融合图像。通过实验结果的对比分析，该算法可以更有效地反映源图像中的特征，融合效果有了明显改善。%For the fusion of infrared and visible images, an improved image fusion algorithm based on the sec- ond generation Curvelet transform is proposed. First, the Curvelet transform was performed on the two source images to obtain the transform coefficients on different scales and in various directions. For low frequency coefficients, ac- cording to the different imaging characteristics of infrared and visible images, an adaptive fusion strategy based on the local statistical properties was employed; for high frequency coefficients on different scales and in various directions, a program for coefficient selection based on local-region energy matching was employed. Finally, the fusion image was obtained through the inverse Curvelet transform. Through comparative analysis of the experimental results, the proposed algorithm can effectively reflect the characteristics of the original images, and the fusion effect is improved significantly.