Advanced Real-Time Video Dehazing and Smoke Reduction Algorithm for Indoor Fire Operations

Abstract

Due to the presence of intense smoke and haze, which severely restricts visibility and situational awareness, inside fire operations provide a significant difficulty for first responders. This research study provides an innovative real-time video dehazing and smoke reduction algorithm created specifically for indoor fire scenarios in answer to this urgent demand. To overcome the difficulties presented by smoke and haze, our approach integrates cutting-edge machine learning methods with computer vision techniques. The fundamental parts of the methodology are examined in this research, including picture acquisition, dehazing methods, transmission map refinement utilizing convolutional neural networks (CNNs), and post-processing. We also go through how Cython is integrated for low-latency processing, provide experimental findings, and consider potential applications in both indoor firefighting and other fields.