人脸图像年龄估计系统的设计与实现

The system design and implementation of age estimation from facial images

人脸图像年龄估计具有重要的理论研究意义和潜在的应用价值。提出的人脸图像年龄估计系统主要包括图像预处理,特征提取,学习模型的建立和模型预测等四个功能模块。系统首先采用灰度变换,几何归一化和直方图均衡等图像预处理技术来得到所需的输入图像,接着选用FG-NET数据集,采用卷积神经网络来进行特征提取,采用支持向量机模型进行学习模型的构建,最后输入新样本到已建立模型中进行年龄段分类和年龄估计。仿真实验表明,系统的正确率为88.14%,年龄总误差为5.3,具有识别率高,应用性强的优点。

Age estimation from face images is of great theoretical research significance and potential applied value. The proposed system of age estimation from face images mainly included four function modules, which were image preprocessing, feature extraction, the establishment of estimation model and model prediction. Firstly,the desired input images were gained via the image preprocessing techniques, including gray-scale transformation, geometric normalization and histogram equalization.Next, the features were extracted using convolutional neural network in the FG-NET database. Then, the model of learning model was constructed using support vector machine classifiter. Finally, the established model can give age classification and estimation results when the new sample image is input into the system. The simulation results show that the accuracy of the system is 88.14% and the total error is 5.3. The proposed system has the advantages of high recognition rate and strong applicability.