Title:	3D FACE RECONSTRUCTION FROM HARD BLENDED EDGES
Author(s):	Yueming Ding and P.Y. Mok
ISBN:	978-989-8704-32-0
Editors:	Yingcai Xiao, Ajith Abraham and Guo Chao Peng
Year:	2021
Edition:	Single
Keywords:	3D Face Reconstruction, Feature Extraction, Deformable Model
Type:	Full
First Page:	21
Last Page:	28
Language:	English
Cover:
Full Contents:	click to dowload
Paper Abstract:	3D face reconstruction from 2D images is an important research topic because it supports a wide range of applications, such as face recognition, animations, games, and AR/VR systems. 3D face reconstruction from contour features is a challenging task, because traditional edge detection algorithms produce a lot of noises, which are prone to making the reconstruction model trapped in a local optimum or even being degraded. With the development of deep learning, a lot of researcher introduce neural network into contour detection, which can extract relatively clear contours compared with previous methods. In this article, we employ a hard blended face contour feature from neural network and canny edge extractor for face reconstruction. Our method not only improves the 3D face model reconstruction accuracy on synthesis images, but performs more accurately and robustly on in-the-wild images under blurriness, makeup, occlusion and ill-illumination conditions.
	Go Back