基于神经网络的人脸检测和优选方法

Face Detection and Optimization Method Based on Neural Network

随着计算机视觉技术的飞速发展,人脸的检测技术已经广泛用于人们的日常生活中。人脸检测技术的改善与进步会使人脸识别系统的效率提高,从而使我们的生活变得更加便捷。论文参考前沿神经网络算法研究成果,在此基础上结合肤色检测、清晰度检测、以及人脸姿态估计构建一个基于视频数据的人脸检测加优选的人脸推送方法。以MTCNN网络结构为基础,针对大多数人脸识别算法对于待识别图片的需求,将最优最好的人脸图像供其识别,从而达到提高人脸识别效率的目的。

With the rapid development of computer vision technology,face detection technology has been widely used in people’s daily life. The improvement and progress of face detection technology will improve the efficiency of face recognition system,and make our life more convenient. On this basis,combined with skin color detection,definition detection,and face pose estimation,a face detection and optimal face push method based on video data is constructed. Based on the MTCNN network structure,the best and best face image is used for recognition of most face recognition algorithms,so as to improve the efficiency of face recognition.