Improved pyramid image fusion based on pseudo inverse

In this paper, the noise of the quantification errors in reconstruction of traditional Laplacian pyramid (LP) transform in introduced. The principle and pattern of the reconstruction in improved Laplacian pyramid transform based on pseudo inverse are addressed. According to the similarities of major kinds of pyramid transforms, a series of improved pyramid transform are proposed for Contrast pyramid, Gradient pyramid and Morphological pyramid through the pseudo inverse reconstruction. And the improved image fusion frameworks are experimented on multi-focus images and the images with Gaussian white noise respectively. Experimental results clearly indicate that the improved pyramid transforms is meaningful in retaining quantification noise.