Multi Spectral Image Fusion with Deep Convolutional Network

摘要

A new multispectral image fusion method is proposed, based on deep convolutional neural networks. For pan-sharpening problem, the proposed method utilize the both super-resolution fusion methods and deep convolutional neural network. By the spatial information from the panchromatic (PAN) image the Multi-Spectral (MS) image is enhanced. In the other hand, the proposed method is independent from the number of MS bands because the spatial information directly estimated from PAN image. Experiments on images of the representative database are shown, proposed method can achieve better result competitive with the current well known methods.