This article presents an algorithm and mathematical model for an identifying flaming combustion automated process by an unmanned aerial vehicle. A digital image processed by software filters serves as a resource about an accident and fire at oil and gas industry facilities. The algorithm deals with the aircraft height, the movement speed relative to the observation point, as well as relative to the surface background, the combustion source brightness and the investigated area flicker frequency. The generated mathematical model allows to detect a fire hazard by a drone without operator intervention.