基于多特征提取和粒子群算法的图像分类

现有图像分类大都采用单一特征,不能利用多个特征之间性能互补优势,且将特征选择与分类器构造分割开来,影响图像分类的精度和分类器的泛化能力。针对以上问题提出一种基于混沌二进制粒子群算法(CBPSO)的特征选择和SVM参数同步优化方法,利用图像的综合特征,将特征选择和SVM分类器构造结合同步优化,仿真实验结果表明,该算法能同步找出最优的特征子集和合适的SVM参数,提高了图像分类精度和分类器泛化能力。

Image Classification Based on Multi-feature and Particle Swarm Optimization

Most of existing image classification adopts single feature, losing complementary advantages between muhiple features and splitting feature selection and the choice of the SVM parameters.This will affect the accuracy of image classification and generaliza- tion ability of classifier.To solve these problems, a simultaneous feature selection and SVM parameter optimization algorithm based on chaotic binary particle swarm optimization（CBPSO） algorithm is put forward.With the help of CBPSO,the feature selection and parame- ters of SVM could be optimized synchronously.Simulation results show that the proposed algorithm can effectively find out the optimal feature subset and appropriate SVM kernel function and parameters,and improve the precision of image classification and generalization ability classifier.