Image Fusion Algorithm Based on Laplacian Pyramid and Principal Component Analysis Transforms

Abstract

Herein, a novel image fusion algorithm based on Laplacian pyramid transform and principal component analysis (PCA) is proposed. In this algorithm, a multi-resolution transform called Laplacian pyramid transform is used to decompose the original image into sub-images. In the transform domain, the coefficients of the various sub-images are combined using a novel fusion rule that utilizes the PCA technique. Finally, the new coefficients are used in the reconstruction of the fused image. Numerical simulations are performed to test the validity and capability of the proposed fusion algorithm. The average gradient values of PCA fusion, Laplacian pyramid fusion, and the proposed fusion method, are tested, respectively. The average gradient values of the proposed method are larger than the other two methods, especially for the spectral bands over 600nm. The three methods performed a bit inferior between the 500nm and 600nm bands compared to the other spectral bands. The Laplacian pyramid transform fusion algorithm seems relatively stable, but it is obvious that the proposed fusion algorithm has a superior performance in the visible range. Moreover, the principal component of the original image was utilized, thus, preserving the spectral information of the hyperspectral image.