Urban and Rural Construction Land Expansion Pattern Recognition and Analysis Based on Multi-Temporal Remote Sensing Image Analysis

The rapid expansion of urban and rural construction land presents increasing challenges for sustainable development, making accurate recognition of spatial and temporal change patterns essential. This study introduces a comprehensive framework for multi-temporal remote sensing image analysis that integrates the Multi-Temporal Pattern Forecaster with advanced strategies for enhancing spatial temporal representation. The Multi-Temporal Pattern Forecaster comprises three core modules: the Constraint-Aware Manifold Optimizer, which preserves spatial relationships under evolving land use conditions; the Event-Driven Spatial Router, which adaptively highlights significant spatial transitions, and the Probabilistic Temporal Decoder, which models temporal evolution to improve the prediction of future land use states. To further strengthen the framework, two complementary strategies are incorporated. Event-Based Feature Extraction identifies meaningful local changes through image differencing and aggregates spatial contextual information to represent key transformation events. Manifold Adaptation Refinement aligns feature manifolds across consecutive time periods by minimizing distribution discrepancies while embedding domain-specific spatial constraints, thereby improving consistency and generalization across diverse landscapes. Experimental results demonstrate that the proposed methodology achieves superior performance in construction land detection and recognition tasks compared to existing approaches. By effectively capturing spatial temporal dependencies and adapting to complex land evolution processes, this framework offers a robust tool for analyzing construction land expansion and supports informed decision making in urban planning and sustainable land management. The framework captures both temporal change information and spatial object representations from multi-temporal imagery.