Research On The Occlusion Positioning Of Forest Fire's Video Surveillance System Based On GIS

With the continuous development of China's forestry information technology,digital video surveillance systems used for forest fire monitoring have been established in most of forestry region.And many of which have the function of automatic identification of forest fires and automatic calculation of fire location.However the positioning method could only solve the condition of the visible ignition point so far.Therefore,finding the geographical position of forest fire when it is blocked by mountains is of great signifieanee for full view positioning of forest fire.Based on digital video surveillance systems established by research group for automatic identification and positioning of forest fire,A solution has been put forward to the problem that the fire point cannot be located blocked by mountains.The solution is combined smoke triangular diffusion model with exhaustive searching.The main contents are as follows:1.With the demand of location technology based on digital video surveillance systems used for forest fire monitoring,ArcMap is used to analyze the occlusion area in forest monitoring area.Then interpolation optimization is used to reduce the negative effect of the open DEM grid map's accuracy limit.Moreover,the forest's three-dimensional map is made by geographic coordinate transformation and map registration based on satellite remote sensing map and DEM grid map.2.Based on triangular diffusion characteristics in the early stage of forest fire,different edge extraction methods are compared and analyzed.Finally,moving region accumulation method based on Horn-Schunck optical flow analysis is used to separate out the smoke diffusion profile.Then the occluded pixel coordinates in the image on the fire is fitted out successftully.The accuracy rate has reached 81.5%under the condition of non HD.It provides the basic information for the conversion of geographic coordinates.3.With the position of the forest fire based on camera calibration technology,spatial imaging model of camera is modified to adapt solution of occlusion demand.Then exhaustive searching is used to confirm the location of the fire point behind the mountain.Finally the experimental error is within 170m.4.With the digital video surveillance systems established by research group for automatic identification and positioning of forest fire,C#programming language is used to improve and realize the 3D positioning system of forest fire.According to the test,the positioning error is within the range of 200m which could match the engineering application requirements.