A super-resolution reconstruction method of remote sensing image based on deep convolutional neural network

Abstract

A method for super-resolution reconstruction of remote sensing images based on deep convolutional neural networks, involving digital image processing. Let the input image be I LR , through the deep residual attention network output image I DRAN , image I DRAN After passing through the edge enhancement network, the image I with the edge details in the output image enhanced EEN , and finally let the image I DRAN with image I EEN Fusion to obtain the final remote sensing image super-resolution reconstruction result I SR . The residual dual attention block is used to realize the transfer between features to ensure the integrity of the information. The LFF sub-module is added to each residual dual attention block, which makes the network have stronger feature expression ability. An attention mechanism is introduced, which fuses channel attention and spatial attention, which can focus on important feature information more effectively. An edge enhancement network is proposed to further enhance the edge detail recovery ability of remote sensing images.

Description

technical field [0001] The invention relates to digital image processing, in particular to a method for super-resolution reconstruction of remote sensing images based on a deep convolutional neural network. Background technique [0002] With the rapid development of modern aerospace technology, remote sensing images are more and more widely used in various fields, such as agricultural and forestry monitoring, military reconnaissance, urban planning, etc., and their resolution requirements are getting higher and higher. However, due to the limitation of hardware conditions and detection distance, there is still much room for improvement in the resolution and clarity of remote sensing images. Considering the high research cost of improving imaging sensors from the physical level and the long iterative development cycle of hardware, the reconstruction of low-resolution remote sensing images into high-resolution images from the algorithm level is becoming one of the current rese...

AI Technical Summary

Problems solved by technology

The document CN110136063A discloses a remote sensing image super-resolution reconstruction method based on generative confrontation network, but the GAN-based remote sensing image super-resolution reconstruction method has limitations such as easy transition and smooth edges of generated images, and unstable training process.

Images