An Application System Of Fire Warning Based On Image Processing

Outdoor fire monitoring is an important issue involving the safety of people's lives and property.Because traditional fire warning methods are not suitable for outdoor scenes,researchers use computer vision and digital image processing technology to apply video surveillance to outdoor fires.Compared with the traditional fire detector such as smoke sensor,video smoke detection can monitor the scene in real time to detect smoke and make decisions quickly.It can guide the fire rescue direction in the first time,with short response time,high sensitivity and wide coverage.Based on the existing image smoke fire warning method,combined with video image processing technology,three innovative points are proposed to realize intelligent fire warning.They are summarized as follows:1.Smoke image sharpening technology for fire source extinguishing and personnel rescue.In the case of fire,a large amount of smoke will be generated,and tiny particles in the smoke will cause scattering of the atmosphere and affect the quality of the image captured by the camera.In order to achieve an effective and fast smoke removal method,the smoke images collected is de-smoothed.Images with high illumination intensity use Retinex theory for image enhancement and the images with small illumination intensity use dark channel prior algorithm for image enhancement.By constructing a fire scenes adaptive de-smoke processing system,rescuers can judge the specific situation of the fire scene based on clear images,and then formulate more effective and targeted rescue strategies,and rescue trapped personnel in time to minimize losses.2.Smoke and fire warning technology based on contextual objects detection.For the problem that it can not be judged that there is fire alarm when the smoke is generated,the contextual objects detection model can effectively obtain the context information of the smoke scenes.Firstly,according to whether the smoke scenes will cause the fire alarm to classify the smoke images,the method establishs a contextual objects detection model,and then use the model to detect the smoke images.The experimental results show that the method improves the accuracy of the fire warning.In the case of the non-fire alarm situation,there will be no false alarms on the smoke scene.3.Fire alarm technology based on image semantic description.The method proposes an early fire warning model combined with a semantic description method.While the site generates smoke,it can intelligently assist the monitoring personnel to determine whether the environment within a certain range around the smoke will cause fire alarm and generate basic semantics,which can help to read and call video images conveniently.The difference between the experiment and the traditional video image smoke detection method is compared in experiments and the results show that the method increases the system's adaptability to monitoring scenes and the reliability of fire warning system.