A Novel Image Change Detection Model and Optimization Algorithm based on Game Theory under the Bounded Rationality Conditions

Abstract

In this paper, we propose novel image change detection model and optimization algorithm based on game theory under the bounded rationality conditions. In the field of remote sensing image application, change detection is of the same area is analyzed by two different periods of remote sensing images, testing to determine the region in a period of change information. To enhance the traditional change detection algorithms we modify the method as the follows. Firstly, we analyze the mathematical forms of the game theory as the co-evolution has the basic characteristics of game theory and the dynamic characteristics of co-evolution through continuous evolution and eventually reaches a stable equilibrium state because of the feature. Later, we discuss the novel mathematical optimization approaches for the change detection based on the proposed game theory prior. Then, we combine independent component analysis and the bounded rationality conditions to finalize the detection algorithm. The independent component analysis is a data processing method appeared in recent years which can be as an extension of PCA and it will be the data transform into each other on the direction of the independence with higher robustness. In the experiment part, we simulate the experiment with the other state-of-the-art algorithms. The result reflects that our algorithm outperforms under various conditions and enhances the detection accuracy from 95.8% to 97.9% which has made the new breakthrough.