在人脸表情识别中，基于局部二值模式（LBP）算子算法与传统的特征提取算法相比具有特征提取准确、精细、光照不变性等优点，但也有直方图维数高、判别能力差、具有冗余信息等缺点。本文提出一种C－LBP算法通过加入中心点到算法中进行特特征提取，能够更有效的提取特征数据。再结合使用稀疏表达分类器实现对特征进行分类和识别。经实验结果表明，与传统LBP算法对比，文中算法用于人脸表情的识别的识别率得到大幅度提高。%In facial expression recognition, the algorithm LBP which based on local binary pattern has the following advantages, such as the accuration of characteristics extraction, fine and illumination invariant comparing to the traditional feature extraction algorithm, but it also has disadvantages such as high dimension histogram, poor discriminant abilitys, much redundant information and so on. The C-LBP operator algorithm that presented in the paper extracts the feature data more efficiently by adding the central point to the algorithm for feature extraction. And using the algorithm made the classification and recognition of characteristics which using sparse expression classifier come true. The experimental results show that, compared with the traditional LBP algorithm, the recognition rate of facial expression recognition is greatly improved.
展开▼
