基于证据理论和支持向量机相融的烟雾图像检测算法

针对传统烟雾图像检测算法低检测率缺陷,提出一种证据理论和支持向量机相融合的烟雾图像检测算法(DS-SVM).首先分别提取主方向性状、高低频能量比、烟雾面积增长等3类烟雾特征,然后3类单特征的支持向量机检测结果作为D-S理论的独立证据,构造基本概率指派,最后根据决策规则和判决门限获得烟雾图像的最终检测结果.仿真结果表明,相对于传统检测算法,DS-SVM有效提高了烟雾图像检测率,可以满足不同环境下烟雾图像检测要求.

A smoke image detection algorithm based on evidence theory and support vector machine

The traditional smoke image detection algorithm has flow accuracy, this paper puts forward a smoke image detection algorithm based on multi-feature fusion and support vector machine. Firstly, the principal direction character, high frequency energy ratio and smoke area growth rate are extracted from smoke suspected regional, and then the recognition results of the three single features of support vector machine is taken as independent evidence, and the basic probability assignment is constructed, and finally the D-S evidence combination rules is used to fuse the decision level, and smoke image detection results are got according to the classification decision threshold. The simulation results show that the proposed method can effectively improve the detection rate of smoke image compared with the traditional methods, which can meet the different requirements under the environment of smoke image detection.