【摘要】 基于人脸特征提取问题可以转化为组合优化问题这一思路,提出了仿生算法与主成分分析相融合的人脸识别算法.该方法先通过主成分分析方法得到人脸特征子空间;然后在已有特征的基础上,分别通过遗传算法与离散粒子群算法进一步提取出可使识别正确率达到最高的人脸图像特征.在ORL人脸库上的实验结果表明:与传统的主成分分析相比,该方法不仅能进一步降低特征子空间的维数,从而提高识别速度,而且能获得更高的识别率.更多还原

【Abstract】 Considering that the face feature extraction problem can be transformed into a combinatorial optimization one,a new face recognition algorithm is proposed to integrate the bionic algorithm with principal comment analysis.In the new method,the feature subspace of face pictures is obtained through conventional principle component analysis.Then,the most accurate recognition is given from the subspace by use of the genetic algorithm and binary particle swarm algorithm separately.The experimental results based on ORL face database showed that the new algorithm is able to reduce the number of dimensions of feature space with highly accurate recognition rate.更多还原