摘要
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,本文提出的MB-LBP算法在原来LBP算法的优点基础上能通过分块更有效的提取特征数据。并结合基于压缩感知理论实现人脸表情的识别。本文主要利用LBP算法提取人脸表情特征然后用稀疏表达分类器实现对特征进行分类和识别,经实验结果表明能有效的提高识别率和识别效率。
Compared with the conventional feature extraction algorithms,the algorithm based on Local Binary Pattem（LBP） operator is much more accurate,particular, and has the advantages of illumination invariance.The MB-LBP operator algorithm that presented in the paper not only has all the advantages of original LBP algorithm but also is more efficient than the originals.It combines to the theory of sparse representation can achieve facial expression recognition.In this paper ,by using the LBP algorithm the system can extract facial expression feature and then identify our expression via classifier based on sparse representation theory.The experimental results show that it can improve the recognition rate and efficiency.
出处
《自动化与仪器仪表》
2014年第10期137-139,142,共4页
Automation & Instrumentation
