3D Face Modeling Algorithm for Film and Television Animation Based on Lightweight Convolutional Neural Network

Through the analysis of facial feature extraction technology, this paper designs a lightweight convolutional neural network (LW-CNN). The LW-CNN model adopts separable convolution structure, which can propose more accurate features with fewer parameters and extract 3D points human face. In order to enhance accuracy extraction, face detection method based on inverted triangle structure is used detect frame images in training set before extracts features. Aiming at problem that algorithm difference criterion cannot effectively discriminative information, Generalized Multiple Maximum Dispersion Difference Criterion (GMMSD) corresponding are proposed. uses instead entropy avoid “small sample” problem, use QR decomposition effective for recognition, while also reducing computational complexity extraction. Compared traditional methods, GMMSD avoids samples” does not require preprocessing steps samples; it from original samples retains distribution characteristics samples. According different change matrices, evolve into algorithms, shows generalized GMMSD. Experiments show identification improve recognition.