一种改进的遗传算法在疲劳断口图像鉴别中的应用

A Modified Genetic Algorithm Applied in Fatigue Fracture Image Identification

特征选择是疲劳断口图像鉴别中的重要问题之一。针对高维的断口图像特征,特征选择一方面可以提高分类精度和效率,另一方面可以确定富含信息的特征子集。针对此问题,提出一种改进的遗传算法,并用于疲劳断口的特征选择:首先,引入线性预测对简单遗传算法SGA(simple genetic algorithm)进行修正,在每一代较优的个体附近,采用线性预测的方法预测出新的个体,提高遗传算法的局部搜索能力;其次,把特征选择转化为目标优化问题,确定目标函数,并利用改进的遗传算法进行断口图像特征选择,得到精简的特征子集;最后,在特征子集上进行断口图像分类鉴别。结果表明:改进的遗传算法可找到有效的特征子集,从而实现降维并提高分类精度。

Feature selection is one of the most important problems in fatigue fracture image identification.For high dimension data,feature selection can not only improve the accuracy and efficiency of classification,but also discover informative subset.This paper proposed an improved genetic algorithm and applied it to fatigue fracture image identification.First,genetic algorithm was improved by use of linear prediction to enhance genetic algorithm local search capabilities.Then,the feature selection was viewed as an optimization problem and the objective function was determined,and feature subset was obtained by using improved genetic algorithm.Finally,fatigue fracture image identification was performed based on feature subset.Identification results show that the improved genetic algorithm can find an effective feature subset,achieve dimension reduction and improve the classification accuracy.