【摘要】 提出了一种基于遗传算法寻优特性的小波域加权特征脸识别算法。首先利用小波变换对人脸图像进行一级小波分解得到人脸图像的近似分量和高频分量;其次,使用遗传算法对这2种分量进行了特征选择;最后将选择得到的近似分量和高频平均分量并行进行特征脸识别,并对识别结果进行加权排序以实现人脸识别。在ORL人脸库上的测试结果表明,该算法的识别率高于传统PCA方法和基于SGA的频谱脸识别方法,具有良好的稳定性和鲁棒性。更多还原

【Abstract】 A weighted eigenface recognition algorithm based on optimization characteristic of genetic algorithm in wavelet domain is proposed.Firstly,each face image is decomposed into four sub-bands using one-level DWT(Discrete wavelet transform).Secondly,SGA(Simple genetic algorithm) is utilized into four sub-bands for feature selection.Then,the selected four sub-bands features are processed in parallel using eigenface recognition.Finally,the face recognition result is obtained through weighted ordination.Experimental result in ORL face library shows that the algorithm proposed in this paper has higher recognition rate than PCA and the spectrum face recognition based on SGA.更多还原