Video Smoke Flame Detection Using Spatial Turbulence Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional fire detection methods, such as particle sampling and temperature sensors, are ineffective in quickly detecting the presence of smoke or flame due to location constraints and lack of data on fire size, location, and intensity, while existing video content analysis algorithms often result in false alarms or fail to detect both smoke and flame effectively.
Innovation Solution
A method for detecting flame or smoke in video input by analyzing spatial features, including turbulence and self-similarity characteristics, using a video recognition system that captures and processes video data to identify regions, measure spatial values, and determine the presence of smoke or flame based on these features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional particle sampling detectors are used, then detection accuracy is improved, but detection speed and early warning capability deteriorate due to smoke reaching time delays
Solution Approach 1:
The patent introduces video cameras as an intermediary detection medium that captures smoke and flame visually from a distance, allowing the smoke detection function to be performed without requiring physical proximity or direct smoke contact with the sensor, thus resolving the time delay issue while maintaining detection accuracy
Solution Approach 2:
The patent replaces the mechanical particle sampling system with an optical vision-based system that uses video imaging and image processing algorithms to detect smoke and flame characteristics, eliminating the need for physical smoke transport to the detector
2Reliability
If traditional temperature sensors are used, then detection reliability is improved, but detection range and early warning capability deteriorate due to proximity requirements
Solution Approach 1:
The patent uses video cameras as an intermediary that can detect thermal radiation and visual characteristics of fire from a distance, replacing the need for direct thermal contact with temperature sensors and enabling reliable detection without proximity constraints
Solution Approach 2:
The patent changes the detection parameter from direct temperature measurement to visual and optical characteristic analysis of smoke and flame, allowing detection at longer distances while maintaining reliability through image processing and pattern recognition
3Length of stationary object
If video content analysis algorithms are used, then detection range is improved, but detection accuracy deteriorates due to false alarms and inability to detect both smoke and flame
Solution Approach 1:
The patent segments the fire detection task into distinct smoke detection and flame detection modules, each optimized for their respective characteristics, and combines their results to achieve both broad detection range and high accuracy without false alarms
Solution Approach 2:
The patent creates a universal video-based detection system that can simultaneously detect both smoke and flame by processing visual information through multiple analysis pathways, enabling one system to perform multiple detection functions with high accuracy
4Device complexity
If traditional detection methods are used, then device simplicity is maintained, but information completeness deteriorates due to lack of data on fire size, location, and intensity
Solution Approach 1:
The patent adds spatial and temporal dimensions to fire detection by using video imaging that captures not only presence/absence but also location, size, shape, movement patterns, and intensity variations over time, providing comprehensive fire information while maintaining relatively simple video camera hardware
Data Source
AI summary
A method for detecting flame and smoke using spatial analysis of video input (40) provided by a video detector. The video input (40) consists of a number of individual frames, wherein analysis is performed on each individual frame to detect and outline regions that may contain smoke or flame (42). Based on the defined outline or boundary of each detected region, spatial features associated with the region are extracted (52), such as perimeter/area and surface area/volume. The extracted spatial features are related to one another (54) to determine the likelihood that the region contains smoke or flame. Extracted spatial features may be related to one another using a power law relationship that provides an indication of the turbulence associated with a bounded region, wherein turbulence is a characteristic of both flame and smoke.