基于运动特征分析的森林火灾烟云实时检测方法

本文提出了一种基于地面视频检测系统中烟云序列图像识别的新方法。利用视频图像的灰度信息,通过运动趋势的估计和相似性的测算,有效提高了具有非结构化及随机性特征的烟云图像检测的可靠性。在仿真试验中,使用此方法对美国Alaska州2005年5月1日拍摄的森林火灾序列图像进行了分析,结果表明它对早期产生的烟云区域能够实时检测和识别,并具有较好的鲁棒性。

Forest Fire Smoke Real-time Detection Based on Motion-character Analysis

In this paper we present a new method based on ground video detecting system for early smoke source identification through the real time processing of images. This method is based on gray-level image-sequence information. By estimating the motion tendency and measuring the similarities of regions, it can effectively improves the detection reliability of fire smoke image-sequence, whose target is unstructured and random. In the experiments, we use this method to analyze a sequence of early smoke images about the forest fire of Alaska, which is captured on May 1st, 2005. The results have shown that this method can real timely detect early smoke region, and it is robust.