基于三角扩散模型的遮挡林火图像中火源点确定  

Determination of Fire Points under Occlusion in A Triangular Diffusion Model

作　　者：赵齐[1] 韩宁[1] 罗琴娟[1] 

机构地区：[1]北京林业大学工学院,北京100083

出　　处：《林业机械与木工设备》2016年第9期13-16,共4页Forestry Machinery & Woodworking Equipment

摘　　要：针对目前森林火灾数字视频监控系统中火源点被遮挡情况下无法定位的问题,提出了一种基于林火烟雾三角扩散模型的方法,用于确定林火报警图像上被遮挡火源点的像素坐标。该方法首先对已被识别的林火烟雾图像进行分割和边缘提取,进而运用光流法分析得到林火烟雾运动趋势,在此基础上构建三角扩散模型,从而估算出图像中遮挡情况下的林火火源点像素坐标。实验结果表明,在烟雾遮挡但能够呈现出三角轮廓时,其像素误差在20个像素点内,这一结论可为实现林火数字视频监控系统的全视域定位提供理论支持。In order to solve the problem of the location of fire points under occlusion in a digital forest fire video monitoring system, a method is presented based on triangular diffusion model forest fire smoke to determine the pixel coordinates of fire points on a forest fire alarm image under occlusion. First,this method involves segmentation and edge extraction of an identified forest fire smoke image,followed by forest fire smoke movement trend obtained through analysis with an optical flow method,based on which a triangular diffusion model is constructed, based on which the pixel coordinates of fire points on a forest fire alarm image under occlusion are estimated. The experimental result shows that smoke occlusion can reveal a triangular contour,the pixel error is within 20 pixel points,and this conclusion can provide theoretical support for realizing the full - view positioning of a forest fire digital video monitoring system.

关 键 词：林火定位 边缘提取 光流分析 三角扩散模型 

分 类 号：S762.3[农业科学—森林保护学]