Method for fire identification through infrared and visible-light video image fusion

Abstract

The invention relates to a method for fire identification through infrared and visible-light video image fusion. A fire is identified through an infrared image preprocessing module, a suspicious infrared image judging module, an infrared image flame determining module, and a visible light fusion comprehensive judgment module. The infrared image preprocessing module is used for carrying out graying treatment, region connection, binarization and dilation operation on an image. The suspicious infrared image judging module is used for calculating the difference between the gray-scale map and the binary image to determine a suspicious image of which the scene changes. The infrared image flame determining module is used for calculating the similarity between two sequential frames of suspicious image to determine whether there is flame. The visible light fusion comprehensive judgment module is used for judging whether the RGB value of a corresponding visible-light image meets a color model of flame based on the model of the flame in an RGB color space and finally judging whether there is a fire. Fusion identification of infrared and visible-light images is realized, missing and false fire reporting is avoided, and the accuracy of fire identification is improved.

Description

technical field [0001] The invention belongs to the technical field of production safety monitoring and relates to a method for merging infrared and visible light video images to identify fires. Background technique [0002] At present, there are many technologies and means to prevent fires, among which smoke alarms and ultraviolet flame detectors are widely used. Smoke alarms are suitable for indoor smoke monitoring and early warning, but not for video surveillance and outdoor use. Ultraviolet flame detectors detect high-temperature fires caused by wavelengths below 200-300nm. They are used to detect flames caused by rapid combustion of hot oil. They are very effective for monitoring flames when instantaneous strong light is emitted. The angle of the device monitoring scene is narrow, and the false alarm rate is high under strong sunlight. Therefore, both fire detection instruments have limitations. The use of infrared technology for fire detection has been explored in r...

AI Technical Summary

Problems solved by technology

The patent and non-patent literature disclose that its infrared image recognition fire is realized in the laboratory. From the recognition picture, the flame accounts for at least 30% of the area in the picture. Such an algorithm is suitable for outdoor, complex background or small fire spots at a distance The recognition rate is very low

[0004] If fire monitoring only uses visible light images for recognition, the algorithm design is complex, the recognition takes a long time, and the judgment results are easily affected by external light and background. For example, local strong sunlight may cause false alarms. Not easy to be identified, prone to false positives

If only infrared images are used for identification, there will also be false alarms caused by single information, such as lights are easily misjudged as flame points

Images