Abstract:  In order to improve the accuracy and efficiency of face recognition results of smart access control system, and thus enhance the intelligent service of smart door security, a smart access control face recognition algorithm based on twin neural network is proposed. The wavelet coefficients of the face image signal are obtained by wavelet transform, the appropriate threshold is selected to process the wavelet coefficients, and the inverse transform of the wavelet coefficients is carried out again to obtain the de-noised face image. After the face image is de-noised, the output value of the face image is mapped and processed in the twin neural network to form a feature vector with a dimension of 128. The contrast loss function is introduced to determine the similarity of the face image by comparing the Euclidean distance between the output feature vectors of the sample network, and finally realize intelligent access control face recognition. The experimental results show that the intelligent access control face recognition results and recognition efficiency of the proposed algorithm are significantly better than other algorithms. 

Key words: twin neural network, smart access control, face recognition, image denoising