Abstract: This paper proposes a method of face recognition, which is based on Multi-Threshold Local Binary Pattern(MTLBP). The gray-scale difference is calculated between each pixel and its local neighborhoods of an image. Different thresholds are chosen to code the foregoing gray-scale difference. Face image is divided into multi-regions. Histogram vectors extracted from multi-regions are adopted to describe the human face. Face recognition is completed by blurring the matching results. Experimental results show that the proposed method is robust to expression variations by combining texture and grads information.

Key words: face recognition, Multi-Threshold Local Binary Pattern(MTLBP), texture, grads

摘要： 提出一种基于多阈值局部二值模式(MTLBP)的人脸识别方法。计算图像中每个像素点与其局部邻域点的灰度差，通过选择不同的阈值编码形成MTLBP，采用多区域直方图向量进行人脸特征描述，模糊化多阈值匹配结果进行人脸识别。实验结果表明，该方法能很好地结合人脸的纹理和梯度信息，对表情等变化具有较好的鲁棒性。

关键词: 人脸识别, 多阈值局部二值模式, 纹理, 梯度

CLC Number: 

• TP18