Foundational Theories for Change Detection and Computational Intelligence

This chapter establishes a unified theoretical foundation for remote sensing change detection, comprising a formal problem formulation, a generalized processing pipeline, standardized evaluation protocols, and benchmark datasets spanning SAR, optical, multispectral, very-high-resolution, and heterogeneous modalities. To support the methodologies developed in later chapters, it further introduces key computational intelligence techniques, including representative deep learning architectures such as deep belief networks, autoencoders, and generative adversarial networks, as well as neural architecture search for automated design of task-specific networks.