We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Research on Multimodal 3D Face Recognition Method Based on Deep Learning.
- Authors
Jie Zhang; Chengqing Pan; Jinlin Huang
- Abstract
In order to improve the effectiveness and accuracy of multimodal 3D face recognition, this paper proposes a multimodal 3D face recognition method based on deep learning. Firstly, a 3D facial dataset is selected; secondly, noise is removed from the original multimodal 3D facial image by filling in holes, denoising, and removing sharp points. The denoised results in each pixel have a value between 0 and 1. Then, the multimodal 3D facial image dataset is trained using the multimodal fusion network of convolutional autoencoder in deep learning methods to achieve multimodal 3D facial image fusion; finally, mathematical relationships are used to divide facial regions, and DSC descriptors are used to extract contextual features of the 3D model shape. Based on this, facial similarity is calculated to achieve multimodal 3D face recognition based on deep learning. The results show that the algorithm proposed in this paper outperforms the other two algorithms in all cases, regardless of the threshold value. The error recognition rate is low, and the face recognition time is less than 9.6 seconds. This indicates that the method proposed in this paper can effectively improve multimodal 3D face recognition in efficiency and accuracy, with strong application performance.
- Subjects
HUMAN facial recognition software; DEEP learning; FACE perception; IMAGE fusion; THREE-dimensional imaging
- Publication
Engineering Letters, 2023, Vol 31, Issue 4, p1740
- ISSN
1816-093X
- Publication type
Article