基于PCA与NSCT结合的引导滤波图像融合方法

Image Fusion Method of Guiding Filter Based on PCA and NSCT

针对红外与可见光图像融合中边缘细节信息损失严重以及空间扭曲的问题,论文提出一种基于主成分分析(PCA)和非下采样轮廓波(NSCT)变换相结合的引导滤波图像融合方法。首先,利用PCA变换与加权最小二乘滤波分别提取红外图像的第一主成分信息与可见光图像的空间细节信息,并将结果注入第一主成分信息中;其次,对其结果与可见光图像分别进行NSCT变换,并对高低频信息分别采用引导滤波和局部拉普拉斯能量法进行融合;最后,将高低频图像先进行NSCT逆变换,再进行PCA逆变换,得到融合图像。实验结果表明:融合后的图像纹理较清晰,图像偏暗区域的纹理细节信息得到增强,并且在五个客观评价指标上均优于传统图像融合方法。

Aiming at the problems of severe loss of edge details and spatial distortion in infrared and visible image fusion,in this paper,an image fusion method based on Principal Component Analysis(PCA)and non-subsampled contourlet transform(NSCT)is proposed.Firstly,the first principal component information of the infrared image and the spatial detail information of the visible image is extracted by PCA transform and weighted least square filter respectively,and the results are injected into the first principal component information,the results are transformed by NSCT with the visible images,and the high and low frequency imag-es are fused by the guided filtering and the local Pierre-Simon Laplace energy method respectively,then PCA inverse transform is used to get the fused image.The experimental results show that the fused image texture is clear,the texture details of the dark region are enhanced,and the fusion method is better than the traditional image fusion method in 5 objective evaluation indexes.