图像处理技术在火焰目标提取中的应用

A pick-up algorithm of flame object in fire identification system

在火灾识别系统中,为了精确地提取火焰目标,本文提出了一种综合有效的算法,该算法采用二维最大熵自动阈值法对火灾图像进行了分割处理,随后采用区域生长法,检测出火灾图像中的可疑区域,并以区域生长后的目标区域为模板,采用Otsu阈值选取法对当前图像进行精确分割,提取准确的火焰目标。实验证明该算法能够有效地提取出连续图像序列中的火焰目标图像。

An effective algorithm is put forward in this paper for picking up flame object precisely in a fire identification system. Two-dimensional maximum-entropy automatic thresholding is used for segmentation of the fire image, then Otsu thresholding is used for precise segmentation of the image, after that we can get an exact flame object. Experiments proved that the method can pick up the image of flame effectively from a sequence of images.