基于分块字典学习的稀疏表示人脸识别方法

Face recognition algorithm based on sparse representation of block dictionary learning

字典学习能加强样本字典的稀疏性,得到的训练样本基类可以提高识别速度和精确度,但是对大量特征维数较高的训练样本使用字典学习进行稀疏表示分类运算量非常大。针对此问题,提出一种基于分块字典学习的稀疏表示人脸识别方法。首先将训练样本字典进行分块,使用Metaface字典学习方法对每块样本进行学习得到训练样本基,然后对字典基进行稀疏表示分类,采用投票方式对每块的最小重构误差进行加权投票确定分类结果。在Extended Yale B、ORL人脸数据库上通过实验对比现有方法,结果表明,该方法在训练样本有光照、表情变化的情况下有较高的识别率和鲁棒性。

Dictionary learning can enhance the sparseness of sample dictionary,and the obtained base class from training samples even can improve the recognizing speed and accuracy,however,it is really a large calculating amount by using dictionary learning to sparse representation based classification of many training samples with high feature dimension.To solve this problem,a face recognition algorithm based on sparse representation of block dictionary learning is proposed.First,to divide training sample dictionary and obtain base class of training samples to each block sample by using Metaface dictionary method,then dictionary base sparse representation based classification and determine the classification result by weight voting to the minimum reconstruction error of each block.Since experiments have been compared with existed methods on Extended Yale B,ORL face database,the results show that the proposed method can achieve higher recognition rates and robustness in the case of illumination and expression changes.