Development and evaluation of algorithms and methodologies of an automated spatial change information extraction system from remotely sensed imagery

This dissertation research is designed to develop and evaluate the operational methodologies and algorithms of an automated spatial change information extraction system focused on automated multitemporal and multisensor image registration and neural networks-based change detection. This study has three major interrelated components. In the first component, the impact of misregistration on the accuracy of remotely sensed land cover change detection is quantitatively investigated using Thematic Mapper (TM) images. Band sensitivity, spatial frequency sensitivity, temporal sensitivity of change detection to misregistration are evaluated using semivariograms. The impact of misregistration on change detection accuracy is then evaluated using the proposed Ellipsoidal Change Detection technique.; In the second component, a new feature-based approach to automated multitemporal and multisensor image registration is presented. The characteristics of this technique is that it combines moment invariant shape descriptors and chain code correlation to establish the correspondences between regions in two images. Feature matching is done in both feature space and image space based on moment invariant distance and chain code correlation. The algorithm is accurate, robust, and fully automated. Experimental results using multitemporal TM imagery are presented.; In the third component, the methodologies and algorithms for an automated change detection system utilizing neural networks are developed and implemented. We first investigate the suitability of the application of neural networks in automated change detection using TM imagery and its related network design problems unique to change detection system. We then develop a neural networks-based change detection system using backpropagation training algorithm. This trained network is then able to efficiently detect land cover changes and provides complete information about the nature of change from satellite imagery using the network as a feed-forward network to detect land cover changes in real-time and in large-scale applications.; Our proposed approaches to change feature extraction are unique in several ways. First, the proposed system provides a powerful and reliable tool to effectively integrate various data sets. Second, this system is fully automated. Third, artificial neural network is used for the first time to develop an automated change detection system with complete categorical land cover change information. Its implication on improving the efficiency and accuracy of the change feature extraction and quantification at all levels of applications ranging from local to global in scale is immense.