Fire Detection Algorithms Based On Infrared Video

Fire is a common and frequent natural disaster in human society, its frequent occurrence makes the human life and property suffer big losses. The video fire detection methods based on surveillance system and Image processing technology can effectively solve the fire alarm failure of traditional fire detectors in large space, and will have a broad application prospect.The principle and characteristic of the fire detection technology were summarized, and the fire detection methods based on infrared video were emphatically discussed. A new fire detection algorithm based on rough set and support vector machine was proposed after the study of existing fire detection algorithms. Firstly, the fire video obtained by camera were separated to form images sequence, and the fire of infrared images sequence were segmented by the combining method of background difference and C-V model, so the flame suspected areas were obtained; Then the circularity, the amount of sharp angles, area variance rate and related coefficient in the suspected region were calculated and analyzed, on this basis, a new feature based on the flame edge jitter characteristic was put forward. At last, the attribute reduction algorithm of rough set was used to reduce the characteristic vectors formed by the flame criterion, remove redundant information, lower sample dimension. The support vector machine classifier was established, and the reduced characteristic vectors were sent as input into the support vector machine classifier to classify and recognize. The experimental results show that to make the rough set as a front end system of the support vector machine, the training and recognition rate of support vector machine was speeded up under the premise of fire identification accuracy, and the data processing ability of support vector machine was improved.