MSBRL: A novel approach for fire origin localization via window brightness pattern analysis in surveillance videos

Abstract

Fire origin determination plays a crucial role in forensic investigations, yet remains challenging in complex indoor environments where traditional methods based on burn patterns or chemical residues often face significant limitations or interference. This paper introduces Multi-Source Brightness Ratio Localization (MSBRL), an innovative non-destructive approach that determines fire origins by analyzing relative brightness patterns observed through multiple windows in surveillance videos. By leveraging the unique spatial-temporal characteristics of early fire development, MSBRL provides critical insights even when direct visualization of the fire is not possible. We establish both theoretical and practical foundations for this method: a comprehensive mathematical model incorporating primary and secondary light reflections demonstrates how brightness ratios uniquely determine source locations, while controlled experiments in a real-world fire case validate its effectiveness. Results show that MSBRL can precisely locate fire origins through systematic post-incident validation, successfully replicating observed brightness patterns using controlled light sources. This method’s ability to extract quantifiable data from surveillance footage while preserving the fire scene advances forensic fire investigation capabilities, particularly in scenarios where conventional techniques are compromised by firefighting activities or environmental factors.