Video Recognition Of The Fire Disaster And The Remote Intelligent Alarm System

Now,security monitoring technology has become more and more popular. There are various types of cameras all over the streets. The normal video security monitoring systems include,closed-circuit monitoring system, anti-theft alarm system, parking management system, electronic fences, patrol system, electronic examination system and so on. How to make full use of video monitoring systems to help with fire protection is the future direction of security monitoring development. The products that can alarm fire disaster remotely without human interaction and enable monitoring via mobile terminals are at its initial stage in market.This thesis presents image recognition algorithms of fire characteristics through extracting static and dynamic characteristics of flame based on OpenCV library,such as color, shape and flickering.We apply the algorithm in a real video monitoring system and realize the We apply the algorithm in a real video monitoring system and realizes the remote fire alarm function.The procedure is summarized as the following:(1)transform the real-time RGB image to YUV422 image and calculate the Y component to detect whether it’s day or night;(2)transform the YUV422 image to RGB565 image to compress the image and extract the red component;(3)perform jitter detection on the red component image to get jitter frequency of suspicious flame area and compare the frequency with the threshold;(4)perform color detection on YUV space to get suspicious area;(5)convert the two image above to greyscale image and merge them together;(6)apply further processing on the merged image to get the final detection results. At last, if there is any fire detected, the system will send short-messaging-service(SMS) message to mobile terminal via third party APIs and show the live picture of the fire via client on mobile terminals. Evaluation results show that the algorithm has good performance and detection accuracy and can achieve remote fire alarm and real time monitoring which has high business value and bright future.