Research On Land Cover Change Detection Method And Implement Of The System Oriented To National Geographic Conditions Monitoring

The global problems caused by the movement of nature, human activities and the continuous development of economic globalization, such as disasters, excessive land reclamation and climate changes make people aware that it is of great significance for making the geographical conditions monitoring put into positive effect and implementing the sustainable development in our country to carry out change detection for land cover rapidly in a large scale the rapid and analyze its characteristics, causes and effects. Remote sensing technology reflecting quickly the real situation of the earth’s surface without direct connect has the advantages of convenience and repeatability of the acquisition of remote sensing image, so it is reliable for the geographical conditions monitoring to use multi-scale, multitemporal and multi-source remote sensing data to detect changes for land cover dynamically.This paper summarized some research achievements made by predecessors and proposed the change detection method used high resolution remote sensing images by integration of multi-source data including vector data and raster images. The main research contents of this article are shown as follows:First, to study the change detection method using high resolution remote sensing images and make the experiment applying the multivariate change detection based on Canonical Correlation Analysis(CCA). The experiment shows that this method on the basis of CCA has a low precision because there are some irregular features which do not perfectly match with the boundary of the change object. At the same time, the change detection method only using high resolution remote sensing images without the participation of shape parameter leads to some pseudo changes. When the spectral information between the two remote sensing images difference greatly, preliminary range detected is not accurate enough.Second, in order to remedy the defect that there are some irregular features which do not perfectly match with the boundary of the change object in the change detection method based on high- resolution remote sensing images and improve the accuracy of the change detection, the paper researches the change detection method using high-resolution remote sensing images with vector data as an auxiliary which make use of the auxiliary information providing prior knowledge in the vector data. On the one hand, high-resolution remote sensing images are segmented on the basis of the boundaries and categories of features in the vector data which makes the image segments match the goals of "different spectrum among classes" and "same spectrum within classes"; On the other hand, the feature space vectors are built by extracting the spectral, texture and shape features based on the image segments obtained and optimizing these features which avoid data redundancy. The result indicates that the multivariate change detection using high-resolution remote sensing images with vector data as an auxiliary based on Canonical Correlation Analysis(CCA) can effectively remove broken features caused by different spectrum among classes, which provides a more effective way to resolve the salt and pepper noise. And through the more experiments of different regions, this method is proved to be have good robustnessThird, based on the analysis of the change detection methods at home and abroad, in accordance with the procedure about the change detection method integrating vector data into high-resolution remote sensing images, standing on the perspective of the national geographical conditions monitoring, for which the change detection system used land cover providing strong technical support is designed and implemented. There are five modules in this system including the task management module, the map operation module, the interpretation module, the change detection module and the sample database module. It not only can be used to detect land cover changes, reducing workload and improving the efficiency with multi-period high-resolution remote sensing images and the integration of vector data and raster images effectively and efficiently, but also has the some functions assisting to correct features, which greatly enhance the level of automation of change detection under the premise of ensuring the accuracy of the change detection.