Smoke Detection Using Semantic Segmentation and Difference Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional smoke detection systems face challenges in accurately detecting smoke due to environmental factors like varying light conditions and smoke density, which can make smoke imperceptible to human eyes, leading to poor accuracy and efficiency.
Innovation Solution
A smoke detection system that includes a camera, storage unit, and processor, which generates a difference image and inputs it into a semantic segmentation model to produce a smoke confidence map, allowing for analysis of whether a smoke event occurs, adapting to dark or bright environments through specific image processing techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a camera is used to monitor smoke with human eye judgment, then the system structure is simple, but the detection accuracy is poor due to environmental light changes and smoke density variations
Solution Approach 1:
The patent replaces the mechanical human eye detection system with an automated image processing system. The processor executes modules that perform difference image generation, semantic segmentation, and smoke confidence map analysis, substituting biological detection with computational algorithms that can objectively analyze smoke characteristics regardless of environmental lighting conditions.
Solution Approach 2:
The patent introduces an intermediary processing system between the camera and the detection decision. The image processing modules act as intermediaries that transform raw camera images into processed data (difference images, semantic segmentation results, smoke confidence maps), enabling accurate smoke detection by mediating the information between the optical sensor and the detection logic.
2Ease of operation
If human eye detection is used to judge smoke, then the detection method is simple to operate, but the detection efficiency is low due to poor accuracy and frequent misjudgments
Solution Approach 1:
The patent implements a self-service detection system where the processor automatically performs all detection tasks without human intervention. The system self-manages image acquisition, processing, analysis, and smoke event determination through executable modules, eliminating the need for human operators to manually analyze images while significantly improving detection efficiency and consistency.
Solution Approach 2:
The patent incorporates feedback mechanisms through the semantic segmentation model and smoke confidence map analysis. The system continuously processes sequential images, compares them with previous frames, and adjusts detection decisions based on accumulated evidence, providing feedback-driven detection that improves efficiency by reducing false positives and confirming smoke events through multiple verification steps.
3Adaptability or versatility
If conventional image monitoring is used, then the system can operate in various environments, but the reliability is poor when smoke is imperceptible to human eyes under different color temperature changes
Solution Approach 1:
The patent applies parameter changes by detecting smoke based on temporal and spatial characteristics rather than relying on fixed color or brightness thresholds. The difference image generation compares sequential frames to detect changes, while the smoke confidence map analyzes multiple parameters including smoke density, shape stability, and temporal persistence, enabling reliable detection across varying environmental conditions without being constrained by specific color temperature ranges.
Data Source
AI summary
The present disclosure discloses a smoke detection system and a smoke detection method. The smoke detection system includes a camera, a storage unit, and a processor. The camera acquires a current image and a previous image. The storage unit stores a plurality of modules. The processor is coupled with the camera and executes the plurality of modules. The processor generates a difference image based on the current image and the previous image. The processor inputs the current image and the difference image to a semantic segmentation model so that the semantic segmentation model outputs a smoke confidence map. The smoke confidence map is generated based on whether a current environment is a dark environment or a bright environment. The processor analyzes the smoke confidence map to determine whether a smoke event occurs in the current image. Therefore, a reliable smoke detection function can be achieved.