Video flame detecting method based on multi-feature fusion technology

Abstract

The invention provides a video flame detecting method based on a multi-feature fusion technology. The video flame detecting method includes firstly using a cumulative geometrical independent component analysis (C-GICA) method to capture a moving target in combination with a flame color decision model, tracking moving targets in current and historical frames in combination with a multi-target tracking technology based on moving target areas, extracting color features, edge features, circularity degrees and textural features of the targets, inputting the features into a back propagation (BP) neural network, and further detecting flames after the decision of the BP neural network. According to the video flame detecting method, spatial-temporal features of the moving features, color features, textural features and the like are comprehensively applied, the defects of algorithms of existing video flame detecting technologies are overcome, and reliability and applicability of the video flame detecting method are effectively improved.

Description

technical field [0001] The invention belongs to the technical field of fire detection, and mainly relates to a video flame detection method for extracting flame features using digital image processing and pattern recognition methods, and then judging whether there is a fire flame in a monitoring scene. Background technique [0002] At present, temperature detection and smoke detection are the most widely used and mature fire detection technologies, but such traditional contact detectors cannot meet the detection needs of some special occasions, such as complex buildings such as tall and large space buildings and long passage buildings. Such places. In recent years, visual detection technology has developed rapidly. This type of detection technology has the characteristics of fast response, wide detection range, and low environmental pollution. As closed-circuit monitoring systems are widely used in various buildings today, video fire detection technology is also gradually d...

AI Technical Summary

Problems solved by technology

[0004] Chinese patent CN201885804U and Chinese patent CN201844880U belong to the fire detection technology in the infrared band. They use dual-wavelength infrared light to detect flames. The dual wavelengths are 4.3 microns and 5.0 microns respectively. On the basis of the former, a 3.8-micron wavelength infrared sensor is added to improve the detection accuracy of flames; Chinese patent CN202195883U and Chinese patent CN201191222 belong to the fire detection technology in the ultraviolet band, and the system judges whether there is a flame by processing the collected ultraviolet radiation signals; Chinese patent CN202306757U belongs to the multi-band composite detection technology, which uses color cameras and near-infrared cameras to capture color video images and near-infrared video images of the monitoring site to determine the presence or absence of flames; compared with visible light band sensors, infrared and ultraviolet sensors Usually expensive, which limits its widespread use

[0006] Chinese patent CN101106727 adopts a color CCD camera system, adopts the image difference method, uses the flame color template obtained by statistics, and combines the characteristics of fire jumping, flickering, and instability to judge the flame; Chinese patent CN101493980 discloses a method based on multi-feature fusion Video flame detection method, this method is based on the mixed Gaussian model technology to detect moving targets, and combined with the color characteristics of the flame, motion characteristics and flickering characteristics of the flame area to model, and step by step according to the recognition sequence of motion, color, and flickering For flame detection, the above two patents have simple criteria. For objects similar to flame color, dithering at a certain frequency may cause false alarms.

(A new approach to linear filtering and prediction problems.Transactions of the ASME Journal of Basic Engineering, 1960,82(D):35-45) proposed the Kalman filter tracking method. Detection Research [D]. Tsinghua University: Doctoral Dissertation. 2010) Based on it, an improved Kalman tracking method was proposed and applied to the field of video fire detection, but this method is computationally complex and not conducive to real-time fire detection

[0010] Texture analysis is an important research content in the field of computer vision and image understanding, and its application fields are very broad. (A new method of image texture analysis and its application [D]. Shanghai: Fudan University Ph. A new method of texture description: statistical topographic feature texture, the texture of this texture description method is richer and more intuitive, and easy to understand, but it has not been applied to the field of fire detection

Images