Research On Change Detection Method Of Urban Typical Land Features Based On RCVA

With the acceleration of urbanization in China,urban planning and urban land use have become the focus of research in recent years.At present,there are more and more publicly available remote sensing image change detection data sets,and the spatial resolution of satellite remote sensing image data is getting higher and higher,and the image acquisition cycle is getting shorter and shorter.The change detection of typical urban ground feature elements can provide the basis for urban planning,facilitate the inspection of land use,illegal construction and building demolition,and is also of great significance to realize the optimal allocation of land resources and sustainable development.In this paper,the images of WHU building change detection dataset were used,and the change detection method of urban typical ground features based on robust change vector analysis algorithm was adopted.The main works of this paper were as follows:Firstly,an improved RCVA change detection method based on minimum information length and expectation maximization method was proposed to solve the problem of high missing rate of RCVA method.In the process of threshold classification,the optimal threshold value could be determined by the method,and the information that has changed and the information that has not changed could be divided more accurately.Compared with the unimproved method,the accuracy rate and Kappa coefficient were improved by 4.13%and 8%,respectively.The experimental results showed that the MML-EM screening algorithm with finite mixture model could accurately divide the image changing area and unchanged area.Secondly,the improved RCVA change detection method only considers the spectral information of the image,and a robust change vector analysis method with texture was proposed.This method combines the spectral information and spatial structure information of the image,and extracts the texture information from the spatial structure of the image.The fast texture analysis method selected in this paper was similar to the gray level co-occurrence matrix algorithm,which extracts feature information for spatial structure of image in the global scope,thus fully integrating spectral information and spatial information.This article chooses was fast texture analysis method,this method was similar to gray level co-occurrence matrix algorithm,are within the global scope for images to extract the feature information of spatial structure,thus fully mix the spectral information and spatial information,can overcome local object pixel processing method for image space limitation of information processing.Compared with the improved RCVA change detection method,the accuracy and Kappa coefficient of the object-level RCVA change detection method increased by 0.88% and 1.71%,respectively.Compared with the object level CVA method,the accuracy was also improved.The experimental results showed that the RCVA method with texture improved the accuracy of the results to a certain extent,and was more conducive to the detection of small ground object information in the city.Finally,the screening parameters of MML-EM were analyzed to further analyze the results of change detection.The comparison of experimental results shows that object-level RCVA change detection method has the best effect and the highest accuracy for urban ground object change detection.