Research On Spatiotemporal Fusion Algorithm Based On Multi-source High-resolution Satellite Remote Sensing Image

Remote sensing,as a spatial data acquisition method,is an important part of earth observation.However,due to the limitation of current physical performance of remote sensing satellite sensors,the same remote sensing satellite cannot obtain the satellite image data at the same time,which had both high spatial resolution and high temporal resolution.This situation limits the spatio-temporal dynamic monitoring ability of remote sensing data to a certain extent.The spatio-temporal data fusion technology of remote sensing image is one of the important methods to solve this problem.Its main principle is to combine the spatial information of high spatial resolution image with the temporal information of high temporal resolution image to form a remote sensing image with both high spatial resolution and high temporal resolution,so as to improve the quality and frequency of earth observationIn the experimental study,it is found that several widely used spatio-temporal fusion algorithms can not predict well when the land cover type changes,and there is no combination of real image and spatio-temporal fusion algorithm to make a long time series set to verify the accuracy and effectiveness of spatio-temporal fusion algorithms.To solve the above two problems,this paper studies the spatio-temporal fusion algorithm based on GF-1 and MODIS MCD43A4 data.In this paper,two classical spatio-temporal fusion algorithms ESTARFM and FSDAF are tested,their advantages and disadvantages are summarized.On this basis,a more efficient spatio-temporal fusion algorithm IESTARFM is developed.The real image is combined with the improved spatio-temporal fusion algorithm to make a long time series data set to improve the effective earth observation frequency.The main research content of this paper is divided into the following three aspects:Firstly,this paper took GF-1 satellite WFV image data and MODIS MCD43A4 image data as research objects,the exploration of remote sensing spatio-temporal fusion technology combined with Chinese and foreign satellite data is increased.Based on the classical spatio-temporal fusion algorithms ESTARFM and FSDAF,a more efficient spatio-temporal fusion algorithm is developed combining the advantages of the two algorithms.Secondly,this paper used six evaluation indexes in two different scenarios,spring and autumn,to conduct fusion experiments on three spatio-temporal fusion algorithms and compare their fusion results.The experimental results show that the integrated accuracy of the fusion results of the improved spatio-temporal fusion algorithm IESTARFM developed in this paper is 17.6% higher than ESTARFM algorithm and 14.2% higher than FSDAF algorithm.Finally,this paper combined the real image and IESTARFM space-time fusion algorithm to produce a time series set of a region in Beijing-Tianjin-Hebei area.In 2020,a total of 75images were taken in the region throughtout the year,and the effective observations without clouds are 34,accounting for 45.3%.After the spatio-temporal fusion of IESTARFM algorithm,22 hige-definition images were increased,making the effective earth observation images in this area increased from 34 images to 56 images,accounting for 74.6%,and the effective observation frequency increased 29.3%.