Application Of Multi-source Remote Sensing Image Spatio-temporal Fusion Technology In Regional Change Detection

With the rapid development of remote sensing technology,remote sensing image change detection methods have been widely used in many fields such as meteorology,land,agriculture and forestry,geology,oceans,and environmental disasters.Therefore,how to detect changes in a specific area more quickly,efficiently and accurately is of great significance for analyzing and predicting changes that may occur in the future.However,due to the inherent limitations of remote sensing satellite imaging principles,the frequency of repeated observations of medium and high resolution sensors is low,the number of images is small,and the lack of regional images or quality problems seriously restrict the accuracy of change detection.Spatiotemporal fusion technology is an effective means to solve the lack of data and poor quality,but its application in change detection has not been fully verified.How to obtain high-precision change detection results based on spatiotemporal fusion reconstruction images is a crucial issue.The study uses Taiyuan City,Shanxi Province as the research area,Landsat TM,OLI images and MOD09 GA images as the main data sources,Three typical spatiotemporal fusion models are important means of reconstructing high-resolution images.It analyzes the detection results of changes in reconstructed images with different spatiotemporal fusion models,and obtains the changes in Taiyuan from 2000 to 2020.The specific research content and results are as follows:(1)In the study,five annual Landsat image data reconstruction and quality evaluation in Taiyuan City,Shanxi Province were implemented.Three typical spatiotemporal fusion models(SATRFM,ESTARFM,and FSDAF)were used to reconstruct Landsat images in the study area,and the reconstruction quality was evaluated through the existing original images.The results show that the average deviations of the reconstructed images of the three spatiotemporal fusion models of SATRFM,ESTARFM and FSDAF from the original images are about 0.03,0.027,and 0.028,respectively,and the root mean square errors are about 0.03,0.028,and 0.029,respectively.It can be concluded that the reconstructed image quality of the three spatiotemporal fusion models is good,and the reconstruction accuracy of ESTARFM is the highest.(2)In the study,the IR-MAD change detection and accuracy analysis of three spatiotemporal fusion reconstructed images were implemented.The study uses IR-MAD for change detection on remote sensing images reconstructed by three spatiotemporal fusion models from 2000 to 2020,and evaluates the change detection accuracy through the change error confusion matrix.The results show that IR-MAD has high change detection accuracy in area change detection,with the highest total change detection accuracy reaching 92.4%,and the Kappa coefficient up to 0.85.It also illustrates the reliability of using spatio-temporal fusion model to reconstruct images for change detection.(3)The study shows that the total accuracy of change detection using STARFM model reconstructed images for change detection is between 85.2% and 89.2%,and the Kappa coefficient is between 0.71 and 0.79.The total accuracy of change detection for the image reconstructed by the ESTARFM model is between 91.2%-92.4%,and the Kappa coefficient is between 0.82-0.85.The total accuracy of change detection for the image reconstructed by the FSDAF model is between 89%-91.2%,and the Kappa coefficient is between 0.78-0.82.The results show that the total change detection accuracy and Kappa coefficient of the ESTARFM model fusion reconstructed image are higher than the other two fusion models.It also proves that the accuracy of change detection is affected by the accuracy of the spatiotemporal fusion model,and its accuracy increases with the increase of the spatiotemporal fusion accuracy.The results show that the spatiotemporal fusion model can effectively reconstruct missing data images,and has high total change detection accuracy and Kappa coefficient in IR-MAD change detection.In addition,the IR-MAD change detection accuracy is directly proportional to the spatiotemporal fusion model reconstruction image accuracy.The greater the spatiotemporal fusion model reconstruction image accuracy,the greater the IR-MAD change detection accuracy.Finally,the main change from 2000 to 2020 in Taiyuan City is urban construction,which is closely related to the rapid development of society.