基于插值Directionlet变换的图像融合方法

为了凸显区域的细节特征,改善图像的视觉效果,本文通过对Directionlet变换进行插值处理,并结合压缩感知理论提出一种新的图像融合方法。在进行插值Directionlet变换分解以后,对低频分量通过其区域特征及其接近度进行系数融合;对于高频分量,使用改进稀疏化的哈达玛矩阵进行观测,并对观测值进行加权融合,然后对融合后的高频方向子带进行重构;最后通过插值Directionlet逆变换重构图像。数值试验显示,本文提出的融合算法计算复杂度低,融合效果好,能同时得到目标和场景均清晰的融合图像。

Image fusion method based on interpolation Directionlet

This article puts forward a new method of image fusion based on interpolation processing of Directionlet and compression sensing theory to highlight the details of regional characteristics and improve the image visual effect.After interpolation Directionlet transform decomposition,the low-frequency component coefficients are fused using the regional characteristics and Verbal Immediacy;and the high-frequency component coefficients are fused using the improved sparse Hadamard matrix observation weight fusion method.Then,the fused high-frequency component coefficients are reconstructed.Finally,the image is reconstructed by the inverse interpolation Directionlet transform.Experiment results show that this fusion algorithm can get image with clear object and background,and it has low computational complexity and good fusion effect.