The Research of Single-Sample Face Recognition Based on Wavelet Image Fusion

As an important part of robot and human-computer interaction, face recognition technology can improve the service and security of robot. In recent years, face recognition technology has improved a lot. Now researchers mainly pay attention to study multi-pose and multi-sample face recognition, but it's difficult to get the method of obtaining these images. However, it is easy to get single face image of per person. So, it is very significant to study the face recognition with single training sample. This paper introduces the single-sample face recognition research based on wavelet image fusion. Firstly, it uses the methods of wavelet transformation and image fusion to obtain the low frequency information of registered image and deposit it to library. Then fusing the low frequency information in library and the high frequency information in tested images. By computing the Euclidean distances between these two images and take it as the input feature to the nerve network to classify. Also, it will use BP neural network to make up the classifier of face recognition and improve it based on traditional neural network. Single face image will be designed matched to every face, activation functions and nodes of nerve cells of the input layer, hidden layer and output layer are designed actively. Lastly through the experiments on the face detection of FERET base, the result is found that the classifier designed in this article is useful for the detection and recognition of human faces with perspective angles, ornaments and in different sizes.