Change Detection Based On Multitemporal Remote Sensing Image

Multitemporal remote sensing image change detection has been applied in civil economy and national defense widely. By analyzing multi-temporal remote sensing images about same region, change detection is used to provide land change information for updating data of resource and environment, analysising the situation of battlefield and evaluating the affection of attack. On the base of system analysis for change detection, some key theories and approaches in change detection and its applications are researched in this dissertation, which include image registration, radiometric correctionand and some change detection theories and approaches such as image transformation, image difference and structure feature analysis.Multitemporal image registration and radiometric corrections are two important preprocessing methods of the various requirements for change detection, whose accuracy could directly affect the capability of change detection. A multitemporal remote sensing image registration algorithm based on limitation of the gray change scope about feature points and their neighbor between two temporal images is developed in this dissertation. In the algorithm, the accuracy and robust of registration can be improved by robust parameter estimation and hierarchical structure. A relative radiometric correction algorithm based on least absolution deviation regression is also proposed in this dissertation, in which robust estimation and iteration method are used to eliminate the influence of change area.Image transformation based change detection technique is one kind of the most important techniques for multispectral images. Traditional image transformation methods are based on orthodoxy theory, which might be difficult in deal with higher order dependences in multitemporal images. A change detection method is developed in this dissertation, which applies independent component analysis to multi-temporal multi-spectral remote sensing image transformation. The dependences in multitemporal multispectral images by independent component analysis are reduced. In the algorithm, damped factor is imported to reduce the dependence on initial weights, thus the robust of the algorithm is improved.Image difference is the most direct change detection method for change area extraction, but the gray difference image that is only based on spectral feature is difficult in describing the structure change of an objects. An image difference change detection...