基于边界矩和支持向量机的火焰识别算法

Flame recognition algorithm based on boundary moment and support vector machine

根据火焰的燃烧特性,结合火焰的空间形状特征和动态变化特征,设计了一种基于动态边界矩和支持向量机的火焰识别算法。利用相邻帧边界矩不变量的差值来描述火焰的动态特征,基于支持向量机对火焰和疑似火焰目标样本进行分类检测。实验表明,该算法具有较好的火焰目标识别性能、较低的虚警率和较强的抗干扰性能。

This paper analyzed the burning characteristics of flame, designed the flame recognition algorithm with the shape of space and dynamic flame features. It described the dynamic characteristics of flame by the difference of boundary moment invariants between adjacent frames, classified and detected flame and suspected flame targets by support vector machine. The experiments show that the algorithm has better ability to recognize flame targets, lower false alarm rate and stronger anti-jamming ability.