Flame Detection and Recognition Method Based on Infrared Video Image

Abstract

The invention discloses a flame detection and identification method based on an infrared video image. The method firstly uses an infrared camera to collect an infrared video image, obtains an infrared light image of the previous frame and the current frame, and then uses the brightness threshold to extract the adjacent two frame images respectively. the suspected flame area; then perform target matching on the suspected flame area in the adjacent two frames; then perform target deletion judgment on the target matching result, and use the set target confidence and the correlation of the same target in the adjacent two frames to carry out the flame target. identify. The method can effectively and quickly identify the flame target, and can also ensure the recognition of the flame target in the night environment. The dynamic deletion of the target memory during the recognition process can reduce the consumption of computer memory during the algorithm running process.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a flame detection and recognition method based on infrared video images. Background technique [0002] In recent years, the frequent occurrence of large-space building fires, mine fires, forest fires, and tunnel fires not only caused huge losses to human life and property, but also caused great damage to the human ecological environment. With the continuous progress of science, people began to focus on the research of intelligent video analysis to prevent fires, and the method of flame detection and identification based on image and video analysis has been rapidly promoted. [0003] Flame detection and recognition methods based on image and video analysis usually include the extraction of suspected flame areas and flame recognition. The extraction of suspected flame areas is the premise of flame recognition, and the separation of flame images from the background is the b...

AI Technical Summary

Problems solved by technology

Among them, the flame recognition method that only uses static features such as color is susceptible to interference from scenes similar to flame colors, which affects the robustness of the system

Phollips et al. use the intensity of the gray histogram of the flame and the time changes of adjacent frames for flame recognition. They also need a better detection environment (less moving non-flame light interference), and their experimental data are subjectively determined by themselves. , if the input data changes, the effect will be greatly reduced

Yamagishi et al. proposed a flame detection method based on a neural network, and the calculation of the algorithm is relatively large

At the same time, some nights are affected and 24 / 7 monitoring cannot be guaranteed

Images