An early forest fire detection method based on a video image

Abstract

The invention provides an early forest fire detection method based on a video image, which relates to the technical field of video remote monitoring fire. The method includes flame detection and smokedetection. First, the suspected flame region of the forest video image is obtained by flame detection. The effective static and dynamic characteristics of the flame are extracted from the suspected flame area, and the characteristics are fused. Then the forest images are classified by support vector machine to judge whether there is flame or not. Smoke detection first obtains the suspected smog area of the forest video image. The effective static features of smoke are extracted from the suspected smoke area, and the features are fused. Then the support vector machine is used to classify the smoke image to determine whether there is smoke in the collected forest image. The early forest fire detection method based on a video image simultaneously takes into account the identification of smoke and flame, and can detect the fire in the video in real time, eliminate the influence of disturbances in the forest environment, and accurately discover the early fire situation.

Description

technical field [0001] The invention relates to the technical field of video remote fire monitoring, in particular to an early forest fire detection method based on video images. Background technique [0002] Forests play an important role in the survival and development of human beings. Human beings and forests are interdependent, restrict each other, and prosper and develop together. Forest fire ranks first among the three natural disasters that destroy forests. In addition to causing huge economic losses and casualties to human beings, forest fires will also cause great harm to forest resources and the ecological environment. Therefore, using advanced scientific knowledge to develop a set of reliable forest fire detection methods can greatly reduce losses and fire fighting costs. [0003] Traditional detection methods mainly use manual observation, aircraft aerial survey, satellite detection and other methods for fire warning. However, due to problems such as low timeli...

AI Technical Summary

Problems solved by technology

However, due to problems such as low timeliness, poor detection effect when the smoke above the fire is occluded, and economic costs, the above method is not optimistic.

In view of the characteristics of large space, large area, and complex environment such as forests, traditional fire detection technologies (smoke, temperature, and other sensor-based fire detectors) are subject to many limitations.

Images