摘要： 针对已有固定火焰图像特征模式识别算法泛化能力较差,且误报率较高的问题,提出一种新的火焰图像特征自适应选择算法。根据特征约简的2大基本准则,将遗传优化引入到粗糙集的属性约简,使交叉和变异概率随个体的适应度值自适应调整,以保护较优并淘汰适应度值低
的个体。通过动态修剪并补充新个体增加种群的多样性,从而提高遗传算法的全局寻优能力。实验结果表明,与基于支持向量机的图像型火灾探测算法相比,改进算法在降低特征空间维数的同时,火焰的平均识别率提高了16%。
关键词:
遗传算法,
粗糙集,
属性约简,
全局寻优,
适应度函数
Abstract: Aiming at the lower generalization ability and high false rate of the present pattern recognition algorithms with fixed flame image characteristic,the algorithm of adaptive selection flame image features is proposed in this paper.According to the two basic principles of characteristic reduction,genetic optimization is introduced into the attributes reduction of Rough Set(RS).The ratios of crossover and mutation are changed with individual’s fitness to protect good individual and eliminate bad individual.It dynamically clips the similar individuals and adds new individual,increases the diversity of population to improve the global optimization ability of Genetic Algorithm (GA).Experimental results show that the algorithm can reduce the dimension of feature space,and the average recognition rate of the flame is increased by 16% compared with the image fire detection algorithm based on Support Vector
Machine(SVM).
Key words:
Genetic Algorithm(GA),
Rough Set(RS),
attribute reduction,
global optimization,
fitness function
中图分类号: