融合全局和局部特征并基于神经网络的表情识别

在表情中含有最多特征信息的是面部眉毛、眼睛和嘴巴这三个区域,为充分利用这些特征,减少图像中无用信息在识别过程中对计算机内存的占用,提高人脸表情识别系统的准确率和速度,首先采用haar 和 adaboost人脸检测算法,对图像中的人脸进行识别,获得人脸图像并提取眉毛、眼睛和嘴巴,生成局部(眉毛、眼睛、嘴巴)二值化图,利用PCA方法对人脸图像降维,降维后的全局和局部灰度特征值组成一个列向量。样本由表情数据库产生,经过神经网络样本训练后,进行表情识别。结果表明,该系统对人脸表情识别速度明显快于Gabor 小波算法；识别的准确率高于单独使用PCA算法和神经网络算法；消耗内存比用Gabor 小波算法少,运行较流畅。得出结论：因为提取出包含表情特征信息集中区的眉毛、眼睛和嘴巴,尽可能地多保留了这些局部特征信息,因而提高了表情识别准确率,同时,采用PCA方法对原始图像进行降维处理,有效的减少了信息冗余。

Facial Expression Recognition of Fusion of Global and Local Features Based on Neural Networks

There is the most characteristic information in the regions of the eyebrows, eyes and mouth about facial expressions. In order to make full use of these features, reduce the amount of unavailable information in the image and occupation of the memory during the recognition process and improve the accuracy and speed of facial expression recognition. Firstly, Haar and AdaBoost face detection algorithms are used to recognize the human face in the image, get face images, and extract eyebrows, eyes and mouth. Generate the binaryzation of image of the eyebrow, eye and mouth. Getting the image of descending dimension by PCA algorithm and a column vector was composed of image of binaryzation and image of descending dimension. The samples are generated by an expression database and trained by neural network samples for facial expression recognition. The results show that the speed of facial expression recognition is faster than that of Gabor algorithm; The recognition accuracy is higher than that using the PCA algorithm and the neural network algorithm alone;The consumption of memory is less than that of the Gabor algorithm and the operation is smoother. The conclusion is that the local feature information is preserved as much as possible, because the eyebrows, eyes and mouths which contain the facial feature information are extracted, the accuracy of facial expression recognition is improved. At the same time, the PCA algorithm is used for the image of descending dimension and reduce the redundancy of information effectively.