融合LSH和LoG特征的人脸识别

为了解决人脸识别率受光照变化影响较大的问题，提出一种局部敏感直方图（LSH）和高斯-拉普拉斯（LoG）特征相结合的人脸识别方法。首先，提取人脸图像中的LSH光照不变特征以及LoG边缘细节特征，然后通过计算各特征的标准差确定自适应融合权重，将其进行特征级融合来弥补单一使用LSH特征带来的细节损失，并以此构建更为有效的人脸特征样本集，最后使用稀疏表示算法对样本进行分类识别。在PIE和AR人脸库上的实验表明，所提方法能很好地处理光照问题，而且在训练样本较少的情况下，依然能获得较高的识别率。

Face recognition based on fusing LSH and LoG feature

In order to solve the problem that the face recognition is greatly affected by illumination variation, a new face recognition method based on combine Locality Sensitive Histograms（LSH） and Laplacian of Gaussian（LoG） feature is proposed. Firstly, the LSH illumination invariant feature and the LoG edge feature are extracted from the facial image. Then, the adaptive fusion weight is determined by calculating the standard deviation of each feature, and the two features are fused to compensate for the loss of details when singly use LSH feature, and constructs a more effective set of face feature samples. Finally, the Sparse Representation based Classification（SRC） is used to classify. Experiments on the PIE and AR face database show that the proposed method can deal with the illumination problem well and obtain higher recognition rate even in the case of less training samples.