| Considering the limits such as slower reaction time,larger scale layout cost of traditional smoke sensors,temperature sensors and other hardware technology.The video-based fire detection has many advantages compared to the traditional technology,especially detecting scene which is difficult to realize by traditional sensors such as large space and field environment.Fire departments,companies and research institutions around the world are trying to put image processing algorithms into the field of fire detection recently.In this paper,based on the hardware and software platform of infrared and visible video streams,the flame detection algorithm and the hardware of the system are improved on the basis of the first phase of the project.the flame and smoke detection algorithms are described in detail according to the characteristics of flame and smoke imaging and image processing theory.In the flame detection,the pre-judgment is done by segmenting the luminance threshold of the infrared image,and the connectivity domain is acquired by scanning.Then,the color criterion is introduced to the corresponding position in the visible image to complete the flame recognition and alarm algorithm,and the algorithm of eliminating common interference sources is enhanced.The flame circularity and flame angle criterion are introduced to enhance the anti-interference ability of the system.Smoke detection is mainly based on OTSU value to improve the differential algorithm to complete the initial decision to obtain the suspect area,and then after filtering and morphological operations,the introduction of smoke color and wavelet transform comparison criteria to confirm the smoke area.The test of system function and interference source show that the system can be used in the field of intelligent fire detection especially working with the traditional monitoring system.It has a very good market prospects for its fast reaction speed,easier to be integrated into cloud real-time monitoring center and so on. |
PDF Full Text Request