中文摘要：

针对林火烟雾图像检测从图像信息中提取的特征,数据记录多,存储大,消耗运算速率的问题,提出了基于运动路径角的烟雾检测探究,对烟雾特征进行优化。在真实的森林环境下,用烟饼燃烧产生的烟雾模拟森林火灾初期产生的烟雾,用视频监控记录烟雾的运动,并且搜集不同情况下烟雾图像和不同情况下天空中云的图像,应用图像处理技术分别对烟雾和云图像进行处理和运动路径角计算,对比分析烟雾和云的运动路径角范围,烟雾独特的运动路径角可以作为识别烟雾的特征。在处理烟雾数据中,将图像包含的烟雾信息转换为烟雾的代数特征,较烟雾的颜色特征、纹理特征、运动面积等特征的提取,数据量减少,存储变小,耗时少。

英文摘要：

To solve the problems existed in the extraction of the characteristics from the smoke images of forest fire, such as image information redundancy, large storage, consumption of computing speed, this pa- per proposed a new way for forest fire smoke detection that was based on the motion path angle to optimize the characteristics of the smoke. In a real forest environment, smoke was produced by burning tobacco cake to simulate early forest fire smoke. Video surveillance was used to record the motion of smoke. Smoke ima- ges and images of cloud in the sky were collected under different circumstances. The motion path angle was calculated to find out the range of motion path angles of smoke and cloud which could be used as the identi- fication feature of smoke. When dealing with smoke data, smoke image information was converted to alge- braic characteristics of smoke. Compared to the extraction of other features, such as smoke color feature, texture feature,and moving area, the motion path angle based data extraction decreased the data amount and storage,and less time was needed to conduct calculation.