Research On Road Information Change Detection Algorithm Based On Multi Temporal Remote Sensing Image

With the development of remote sensing technology,remote sensing image in the imaging accuracy,imaging cycle and other aspects are constantly improving,which provides a good data support for the relevant practitioners to obtain ground information from remote sensing image.Multi temporal remote sensing image refers to the remote sensing image taken in the same area at different times.Using multi temporal remote sensing image,we can analyze the changes of imaging area,and provide relevant support for disaster detection,data update and other research.As one of the most important features in the geographic information system,road is of great significance in many fields,such as economic development,military command,geographic mapping,transportation and so on.In recent years,the rise of smart city,autonomous driving and other concepts promotes the research of road information,and brings more conditions and ideas for the research of road information.The change detection of road information refers to the observation of the change of road information in different phases in the same area.Research on road information change detection technology based on multi temporal remote sensing image can save the time cost of manual analysis of road change,and provide data support for road disaster detection,road information update and other related fields.In this context,in view of the problem of using multi temporal remote sensing image to detect the change of road data,this paper puts forward the corresponding solutions to some of the problems,and the main research contents are as follows four parts.The first part is the main method of road information change detection based on multi temporal remote sensing image.This paper first summarizes and analyzes the research status at home and abroad,summarizes the current mainstream technology and methods,and on this basis,carries out the technology selection.The second part is the error improvement algorithm design of multi temporal remote sensing image.Affected by the imaging mechanism of remote sensing image,remote sensing image will have some problems such as offset and distortion,which seriously affect the reliability of road data and the accuracy of road data change detection.To solve this problem,this paper proposes an error improvement method based on road binary map and a remote sensing image registration optimization method based on cluster analysis to improve the accuracy of remote sensing image registration and reduce the false changes of road data caused by remote sensing image itself.The third part is the design of vector road information change detection method.In this paper,a coarse to fine vector road information change detection algorithm is designed.Firstly,taking each vector road data as a whole,the Hausdorff distance is improved for the problem of vector road data change detection,and it is used for rough screening of changing roads,and the roads are divided into changing areas and suspected changing areas.Then,taking the road segment as the minimum matching unit,segmented change detection is carried out on the filtered road data to detect the local change of the suspected changed road.The fourth part is the experiment.In this paper,the feasibility of the proposed algorithm is verified by experiments,and the advantages and disadvantages of the algorithm are analyzed.