协同KPCANet模型在人脸识别中的应用

为解决手工提取图像特征过程繁复和参数复杂问题，提出一种基于深度学习的协同KPCANet模型。该算法能够对现场采集到的人脸数据和特征进行提取和分类，通过提取分块直方图特征进行编码协同表示，将测试样本归于残差最小的类中对人脸数据进行识别和运算。实验结果表明，协同KPCANet模型在滤波器数量L1=10时一层卷积层与L2=15时二层卷积层的正确率分别达到99.17%和99.44%。协同KPCANet模型不仅能使运算过程简洁，还能提高识别结果准确度，提升识别效率。

Application of Cooperative KPCANet Model in Face Recognition

In order to solve the complex process and parameters of extracting features manually from image features at this stage， a collaborative KPCANet model based on deep learning is proposed， which can extract and classify face data and features collected on the spot. This algorithm recognizes and calculates the face data by extracting the block histogram features and co-representation coding， kernel principal component analysis network， binary hash coding， calculating the coding residuals， and classifying the test samples into the classes with the smallest residuals. The experimental results show that the correctness of the cooperative KPCANet model in the number of filters L1 = 10 layer convolution layer and L2 = 15 layer convolution layer reaches 99.17% and 99.44%， respectively. Collaborative KPCANet model can not only simplify the operation process， but also improve the accuracy and efficiency of recognition results.