Abstract: Face recognition is the pattern recognition and an active field of artificial intelligence research direction. This paper based on Local Binary Pattern（LBP） operator puts forward Local Directional Pattern（LDP） operator. For each face image region, local directional pattern operators are used to extract feature, each area’s characteristics histogram is calculated and the measure of Chi distance is used to compare the histogram. The experimental results prove that the method has a better recognition rate than Local Binary Pattern on Yale and Yale B face database and  has good robustness of light.

Key words: Local Binary Pattern（LBP）, Local Directional Pattern（LDP）, Chi distance measure, features histogram

摘要： 人脸识别是当今模式识别和人工智能领域的一个活跃研究方向。基于局部二值模式（LBP）算子提出局部定向模式（Local Directional Pattern，LDP）算子。对人脸图像进行分块，采用局部定向模式算子对每块图像进行特征提取并计算每块区域的特征直方图，对特征直方图采用Chi距离测度进行比较识别。实验证明，该方法在Yale人脸数据库和Yale B人脸数据库相比局部二值模式有更好的识别率，说明该方法对光照有良好的鲁棒性。

关键词: 局部二值模式（LBP）, 局部定向模式（LDP）, Chi距离测度, 特征直方图