Automatic Generation of 3D Facial Image Using Artificial Neural Network
Takuma Yamamoto, Koosuke Hattori, Ryo Taguchi, Masahiro Hoguro, Taizo Umezaki · Electronics and Communications in Japan · 2014
SUMMARY The face is one of the most important regions for communication with others and for recognition of persons, and consequently face recognition systems are highly desirable. Such systems use images captured by a camera to recognize persons. However, face images are variable because of external factors such as the position of the body, ambient lighting, facial expression, and so on. This variability decreases recognition accuracy, and therefore facial recognition systems must overcome this problem. Recognition with 3D data solves this problem, but 3D measuring systems are expensive. Therefore, we propose a method that estimates 3D face data from 2D images captured by a camera. The method uses artificial neural networks, which learn the relations between 2D facial images and 3D facial data, as measured by a CCD camera and a laser range finder. Then, the artificial neural networks are able to reconstruct 3D facial data from 2D images. The experimental results show that the proposed method is effective.