Monitoring And Study On Smog Recognition In Video Surveillance Of Forest Fire Prevention

With the increasing emphasis of the State on forest fire prevention, forest fire prevention has become a primary responsibility of the forestry sector in all provinces and cities. The visible light camera video image processing and intelligent video image recognition technologies have been widely applied in forestry, public security, fire control and other fields. Thecamera video image smog recognition mode can effectively enhance early prevention and treatment of forest fire and substitute the artificial video surveillance and recognition tasks, thus, it plays an important role in forest fire prevention.For the spreading characteristic of fire, early prevention can effectively control the forest fire and reduce losses caused by it. A lot of smoke will be generated before fire and then visible flame will appear. Since green plants generally cover a large area of the forest, forming a great many of obstructions, it is difficult to discover the flame in the video surveillance. When the fire is discovered finally, it has caused serious losses. The smog can rise and is permeable, which can be displayed on the video image immediately when fire happens. Same with the thinking way of human, once we see large amount of smog, we will subconsciously think that there may be a fire. Therefore, it has important significance for the study on smog recognition in video surveillance of forest fire prevention.This paper, based on the working environment and research area of the author, realizes the recognition of smog in the video, improves the early warning effect of forest fire prevention, and by combining with the products developed by the company, it realizes dual-mode monitoring of forest fire, namely, monitoring the fire by both visible light smog recognition mode and thermal imaging temperature sensing mode.This paper draws on partial previous smog recognition technologies and also proposes different monitoring methods; it realizes the recognition function of cruising or static PTZ camera for various smog as well as the elimination function for various interferingobjects in different detection environment. The PTZ is controlled by two different methods (static monitoring and dynamic monitoring) to identify the smog area according to the analysis on the smog characteristics, e.g. the color, texture and spreading direction of smog, as well as extraction of different texture degree and control on the scale factor of suspect smog area.Major achievements of this paper are as follows:(1) According to the characteristic of the smog that it can make the background area blurry and losing texture information, single frame monitoring is adopted to compare the suspect smog area and the corresponding scale factor of marginal information extracted in different degree; meanwhile, interference area is divided, which eliminate the interference problem and reduce the false alarm rate. Thus, the PTZ is able to monitor the smog area in cruising condition.(2) According to the slow-moving characteristic of smog, the background updating method is optimized, which eliminates the emergent objects and those moving from top to down from the video image and prevents the false alarm information caused by camera shaking.(3) By the combination of two kinds of algorithms and PTZ control, the PTZ realizes real-time monitoring on smog without affecting the cruising condition and realizes dual-mode alarm.In order to test the accuracy of algorithm, experimental comparison is conducted by each algorithm for videos under different circumstances and different elimination methods adopted for different interfering objects. The test results indicate that this paper has better accuracy for smog recognition in video surveillance of forest fire prevention and has certain anti-interference ability; furthermore, the algorithm and design proposed in this paper are truly realized in the products developed by the author’s company.