A Pansharpening Based on the Non-Subsampled Contourlet Transform and Convolutional Autoencoder: Application to QuickBird Imagery

This paper presents a pansharpening technique based on the non-subsampled contourlet transform (NSCT) and convolutional autoencoder (CAE). NSCT is exceptionally proficient at presenting orientation information capturing internal geometry of objects. First, it’s used to decompose multispectral (MS) panchromatic (PAN) images into high-frequency low-frequency components using same number decomposition levels. Second, CAE network trained generate original PAN from their spatially degraded versions. Low-resolution are then fed estimated high-resolution images. Third, another The result low-pass high-pass final pan-sharpened image accomplished by injecting detailed map spectral bands corresponding bands. proposed method tested QuickBird datasets compared with some existing pan-sharpening techniques. Objective subjective results demonstrate efficiency method.