Abstract: The face recognition algorithm based on the hyper-feature patches model can obtain good identify effect in few or just one training examples.Compared with the other fashionable face recognition algorithm in present，it has much practical use and has solved the difficult problem of the low recognition rate caused by the single training example.The face recognition experiment with the ORL face databases shows that the recognition rate is high even in the single training example.

摘要： 基于超特征块模型的人脸识别算法在训练样本很少，甚至为单个训练样本时，能取得较好的识别效果。相比于目前流行的人脸识别算法，该技术具有更好的实用价值，它解决了人脸识别技术中对单训练样本所导致识别率较低的难题。采用ORL人脸图像库的实验结果表明，该算法在单训练样本的情况下识别率仍然较高。