A Method for Simultaneous Super-resolution and Colorization of Satellite Imagery Based on Multi-task Deep Neural Networks

Abstract

The invention discloses a method for simultaneous super-resolution and coloring of satellite images based on a multi-task deep neural network, belonging to the technical field of image processing. The present invention mainly comprises the following steps: 1, making high-resolution and low-resolution grayscale image block training sets; 2, constructing a multi-task deep neural network for model training; 3, based on the constructed deep network and the produced The training set trains the network model; 4. According to the learned model parameters, a low-resolution grayscale image is input, and the output obtained is a reconstructed high-resolution color image. By combining the deep super-resolution network and coloring network with excellent performance, the present invention not only enhances the details of the satellite image, but also colorizes the grayscale image at the same time so that it automatically generates a realistic color satellite image, and reduces the The steps and time of execution have broad application prospects in grayscale image coloring, satellite remote sensing and telemetry and other fields.

Description

technical field [0001] The invention belongs to the technical field of image processing, and more specifically relates to a method for simultaneous super-resolution and coloring of satellite images based on a multi-task deep neural network. Background technique [0002] As the performance of the network with a single function becomes stronger, people's demand for a network capable of handling complex multi-tasks is also increasing. The traditional approach is to use the output of one network as the input of another network to get the final result. Because this method not only requires manual interaction, but also wastes a lot of time to execute one by one, and also considers whether there are compatibility issues between the two networks, so people have to seek other methods. [0003] Existing networks need to have two very important functions, super-resolution and colorization. In terms of super-resolution, reconstruction techniques can be divided into different types, ma...

AI Technical Summary

Problems solved by technology

[0008] In order to overcome the inability to deal with complex and changeable problems in reality and the possible incompatibility between multi-tasks in the above-mentioned prior art; the present invention proposes a simultaneous super-resolution and coloring of satellite images based on a multi-task deep neural network Method; the present invention not only carries out multi-task learning to two completely different aspects, but also solves the problem of incompatibility between networks, and satisfies complex requirements in reality

Images