Multispectral and panchromatic image fusion method based on dense and jump connection deep convolutional network

Abstract

The invention relates to a multispectral and panchromatic image fusion method based on a dense and jump connection deep convolutional network. The method comprises two parts of model training and image fusion, and is characterized by at the model training stage, firstly, performing the down-sampling on an original clear multispectral image and a panchromatic image to obtain a simulation training image pair; secondly, extracting the characteristics of the simulated multispectral and panchromatic images, fusing the characteristics by utilizing a dense connection network, and reconstructing a high-spatial-resolution multispectral image by utilizing jump connection; and finally, adjusting parameters of the model by using an Adam algorithm; at the image fusion stage, firstly, extracting the features of multispectral and panchromatic images, fusing the features by utilizing a dense connection network, and reconstructing a high-spatial-resolution multispectral image in combination with jump connection, wherein the two feature extraction sub-networks are responsible for extracting the features of the input image pair, and the three dense connection networks are responsible for fusing the features, the jump connection and two transposed convolutions are responsible for reconstructing the high-spatial-resolution multispectral image.

Description

technical field [0001] The invention belongs to the field of remote sensing image processing, and in particular relates to a multispectral and panchromatic image fusion method based on a dense and skip connection deep convolutional network. Background technique [0002] Remote sensing images have two important properties - spectral resolution and spatial resolution. Spectral resolution refers to the minimum wavelength range that the sensor can distinguish when receiving the spectrum of the target radiation. The narrower the wavelength range, the higher the spectral resolution, and the stronger the ability of the sensor to distinguish and identify each band of light in the spectrum, resulting in The more the number of bands, the richer the spectral information of the obtained remote sensing images. Spatial resolution refers to the minimum distance between two adjacent features that can be identified on remote sensing images. The smaller the minimum distance, the higher the s...

AI Technical Summary

Problems solved by technology

[0007] In order to overcome the problem that existing multispectral and panchromatic image fusion methods cannot accurately generate high spatial resolution multi

Images