Research On Urban Change Detection Based On IR-MAD Algorithm

At present,with the rapid development of remote sensing technology,remote sensing image processing especially the continuous improvement of the level of application of remote sensing data is more and more widely in all fields of society,at the same time observation application of detection of remote sensing image change detection techniques are widely used in land cover change monitoring,and the use of natural disaster and its development trend,and the changes of city forest monitoring expansion and a series of people living in.How to detect the change of a region more efficiently,faster and more accurately is very important for future analysis and prediction of the possible transformation of a specific future period.IR-MAD(the iteratively reweighted multivariate alteration detection)algorithm weighted multiple iterative detection is a method for detecting image changes,because of its simple and effective mining in temporal and multi variable and variable image data,can efficiently capture the changes of unstable points,so in the image change detection is widely used to in this way,the detected change for the initial variables measured at two time points on the independent linear affine transformation is the same,these can effectively avoid the impact of the change in the gain and offset of the data acquisition measurement device.But for massive remote sensing image detection,the traditional way of preprocessing through downloading images is the final change detection,which requires higher hardware infrastructure.The process is too tedious,which has great application timeliness and limitations.Therefore,Google Earth Engine online programming system platform has many advantages,such as high efficiency,high expansion and high compatibility.It has great advantages in solving massive data storage and computing problems.This paper is based on multiple detection methods(IR-MAD,the,iteratively iterative weighted reweighted multivariate alteration detection)algorithm,choosing Nanchang city as the study area,with 20 years of data in the image as the data source,the twenty to four time period a total of eight different Landsat image,through the GEE(Google Earth Engine)as the platform,using Python language IR-MADalgorithm,detection of city change of Nanchang in the twenty years in GEE,at the same time,based on the Google Earth Engine data source in Nanchang twenty phase change detection map,visually display the city of Nanchang from 1998 to 2017 twenty years.At the same time detect tool needs by Docker and ArcGIS and ENVI software for the corresponding image processing,the image changes will eventually get,will change detection results obtained images and reference image Google Earth were compared,and the accuracy of assessment,the final conclusion.The main research work is as follows:(1)this paper introduces the basic concepts and processes of multitemporal remote sensing image change detection from the aspect of theory,including image data preprocessing,information extraction and change detection and detection results to assess the accuracy of variation method,and from the angles of the pixel level,feature level and object level based on a brief overview of the development and the research status of change detection technology at present,and the existing detection methods of the advantages and limitations.(2)according to the detection of changes introduced in more and more multivariate change detection for detection of area for efficient and convenient way more and more,in this context,the change detection process this paper introduces a new way to detect changes in Google Earth on the Engine platform in the form of code on the platform of the whole algorithm finally,after the results were finally cut,after image segmentation image change graph density.(3)the innovation of this paper lies in the image change detection method and the image processing software through preprocessing,clipping and correction of different detection methods and the final data change chart of the early data preprocessing and later,all through the Earth Engine cloud platform in Google programming environment,for the future by detecting the changes of a region provides a new way to ensure a certain accuracy at the same time,tedious steps can greatly improve the image processing with the Google Earth Engine will also become after data processing,region detection and method realization,researchers provide more help.