林火烟雾图像识别技术研究

Study on Forest Fire Smoke Image Recognition Technology

摘要: 森林对于人类的生存和发展有着深远的影响，但森林火灾是森林防护的最大威胁，它不仅可以毁灭森林甚至可以威胁到生命安全，因此如何对森林火灾的发生尽早作出检测和警报，已成为各国研究的重要课题，尤其是森林覆盖率高的国家。本文结合统计理论的思想，提出了基于时间序列分析的森林火灾烟雾图像识别的方法，以时间轴为标尺，对森林火灾产生的烟雾进行记录，作出烟雾扩散曲线，该曲线记录了森林火灾烟雾从无到有再到消失的全过程，将此曲线作为烟雾识别的特征。实验证明，该方法易于掌握，计算工作量小，易于应用推广。

Abstract: Forest has a profound impact on human survival and development, but the forest fire is the greatest threat to forest protection, which can not only destroy forest, but also threaten life safety. Therefore how to make an early detection and alert when forest fires happen, has become the important topic of various countries’ research, especially the countries with the forest of high cover rate. According to the statistical theory of thought, forest fire smoke image recognition method is proposed based on time series analysis, which with the time axis as the ruler, records the forest fire smoke, and makes smoke diffusion curve. The curve records the whole process of forest fire smoke from beginning to the disappearance, as the smoke recognition feature. Experimental results show that this method is easy to be applied and popularized due to easy grasp and the less work in calculation.