基于卷积神经网络的时空融合算法

Spatio-Temporal Fusion Algorithm Based on Convolutional Neural Network

摘要: 对于遥感问题，时空融合算法旨在解决卫星传感器无法同时获取高空间分辨率、高空间分辨率的遥感图像的缺陷，大多数情况下，单个卫星传感器的遥感图像无法满足需求，因此衍生出许多融合高空间分辨率、低时间分辨率遥感图像和低空间分辨率、高时间分辨的遥感图像的时空融合方法，其中效果最为显著的便是基于深度学习的时空融合方法，本文在此基础上，利用卷积神经网络构建一种新的时空融合算法，并得到了更小的误差，应用效果比较好，值得应用及推广。

Abstract: For remote sensing, the spatio-temporal fusion algorithm aims to solve the defect that satellite sensors cannot obtain remote sensing images with high spatial resolution and high spatial resolution at the same time. In most cases, the remote sensing images of a single satellite sensor cannot meet the demand, therefore, many methods that merge high spatial resolution images with low temporal resolution and high temporal resolution images with low spatial resolution were created, the most significant effect is the spatiotemporal fusion method based on deep learning. On this basis, this paper constructs a new spatiotemporal fusion algorithm by using convolutional neural network, and obtains less error. The application effect is relatively good, which is worthy of application and promotion.