Fire Image Analysis Based on Spatio-Temporal Fusion Algorithms

Due to the large-scale use of electrical equipment in urban life and the large number of man-made fires and natural fires in the forest environment, according to statistics, the number of deaths from fire in 2018 increased by 0.9% compared with 2017. Under the condition of heavy smoke at the fire scene, it is difficult for firefighters to locate personnel and conduct search and rescue operations under the premise of ensuring personnel's safety. Therefore, we can better get the information of the fire scene and accurately locate the personnel by fusioning the infrared and visible images. Existing fusion methods typically use the same representations and extract the similar characteristics for different source images. However, it may doesn't work for infrared and visible images. In this paper, we use the fusion algorithm named Gradient Transfer Fusion (GTF), based on gradient transfer and total variation (TV) minimization which can keep the thermal radiation and appearance information simultaneously. We formulate the fusion problem as an l1-TV minimization problem, where the data fidelity term and the regularization term keep the main intensity distribution and the gradient variation respectively. To prove the effectiveness of the GTF method, it is compared with other 8 advanced methods and 10 traditional image fusion algorithms from quantitative and qualitative aspects. Finally, the parameter in the GTF method is analyzed and selected, which can be more suitable for image fusion in the fire scene.