Abstract:The increasing frequency and severity of wildfires highlight the need for accurate fire and plume spread models. We introduce an approach that effectively isolates and tracks fire and plume behavior across various spatial and temporal scales and image types, identifying physical phenomena in the system and providing insights useful for developing and validating models. Our method combines image segmentation and graph theory to delineate fire fronts and plume boundaries. We demonstrate that the method effectively distinguishes fires and plumes from visually similar objects. Results demonstrate the successful isolation and tracking of fire and plume dynamics across various image sources, ranging from synoptic-scale ($10^4$-$10^5$ m) satellite images to sub-microscale ($10^0$-$10^1$ m) images captured close to the fire environment. Furthermore, the methodology leverages image inpainting and spatio-temporal dataset generation for use in statistical and machine learning models.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to improve the accuracy in the image segmentation and tracking of multi - scale fire and smoke plume behaviors. With the increase in the frequency and severity of wildfires, there is an increasingly urgent need for accurate fire and smoke plume propagation models. Although existing fire and smoke plume dynamics models have been applied at different scales, they often require a large amount of computing resources, or simplified models are suitable for operating environments but lack the ability to capture fine - grained features. In addition, traditional measuring devices such as anemometers and temperature sensors are difficult to comprehensively capture the complex dynamics of the fire scene because they are costly and can only provide sparse measurement data at a few locations. For this reason, the paper proposes a multi - scale, multi - perspective tool named "Fire Dynamic Vision (FDV)", which can segment and track fires and smoke plumes in visual and infrared videos and statistically interpret their dynamics. Specifically, FDV identifies flame regions, burning regions, and smoke plume boundaries by combining image segmentation techniques in RGB and HSV color spaces and spatial clustering analysis methods. This method can not only effectively distinguish fires and smoke plumes from visually similar objects but also handle videos shot from different perspectives and support the tracking of multiple fire fronts and smoke plumes. The main contributions of the paper include: 1. **Expanding the scope of segmentation and tracking**: including visual and infrared images, not just visual images. 2. **Adding smoke plume segmentation and tracking**: previous work mainly focused on the fire itself. 3. **Supporting multi - perspective video processing**: being able to process videos shot from different angles, not just top - down perspectives. Through these improvements, FDV can successfully isolate and track the dynamics of fires and smoke plumes in a variety of image sources (from satellite images to sub - micron images in near - fire environments), providing important data support for the development and verification of fire models.

Fire Dynamic Vision: Image Segmentation and Tracking for Multi-Scale Fire and Plume Behavior

Level Set Based Flame Segmentation

Multi-Scale Flame Situation Detection Based on Pixel-Level Segmentation of Visual Images

Wildland Fire Detection and Monitoring Using a Drone-Collected RGB/IR Image Dataset

Unsupervised Segmentation of Fire and Smoke from Infra-Red Videos

Fire and Smoke Digital Twin -- A computational framework for modeling fire incident outcomes

A Machine Learning Approach to Waterbody Segmentation in Thermal Infrared Imagery in Support of Tactical Wildfire Mapping

Automatic real-time fire distance, size and power measurement driven by stereo camera and deep learning

California wildfire spread derived using VIIRS satellite observations and an object-based tracking system

Spatio-Temporal Self-Attention Network for Fire Detection and Segmentation in Video Surveillance

A Bayesian Spatio-Temporal Level Set Dynamic Model and Application to Fire Front Propagation

Automated Image-Based Identification and Consistent Classification of Fire Patterns with Quantitative Shape Analysis and Spatial Location Identification

Fire and smoke digital twin – A computational framework for modeling fire incident outcomes

Detection and tracking of gas plumes in LWIR hyperspectral video sequence data

Wildfire Impact Analysis and Spread Dynamics Estimation on Satellite Images Using Deep Learning

Towards a solid solution of real-time fire and flame detection

Physics-driven Fire Modeling from Multi-view Images

Obscured Wildfire Flame Detection By Temporal Analysis of Smoke Patterns Captured by Unmanned Aerial Systems

A High-resolution Large-eddy Simulation Framework for Wildland Fire Predictions using TensorFlow

Using Echo State Networks to Inform Physical Models for Fire Front Propagation

FLAME Diffuser: Wildfire Image Synthesis using Mask Guided Diffusion