Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
NEURAL NETWORK-BASED AUGMENTED REALITY DRAWING METHOD
Document Type and Number:
WIPO Patent Application WO/2021/223133
Kind Code:
A1
Abstract:
A neural network-based augmented reality drawing method. The method mainly comprises six steps: ambient illumination estimation, surface roughness estimation, layer drawing for a virtual object and a plane, optimization of a reflection layer, optimization of a shadow layer, and linear mixing of layers. According to the method, complex reflection and shadow effects are brought into augmented reality drawing for the first time, and the most advanced level at present is reached. The method can be applied to an augmented reality platform, so as to improve the drawing quality and reality sense of the virtual object in a real scene.

Inventors:
ZHOU KUN (CN)
HOU QIMING (CN)
REN ZHONG (CN)
MA SHENGJIE (CN)
SHEN QIAN (CN)
Application Number:
PCT/CN2020/088882
Publication Date:
November 11, 2021
Filing Date:
May 07, 2020
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
UNIV ZHEJIANG (CN)
FACEUNITY TECH CO LTD (CN)
International Classes:
G06T15/50; G06N3/04; G06T15/00; G06T15/04; G06T15/06; G06T15/60
Foreign References:
CN104077802A2014-10-01
CN103761760A2014-04-30
CN101710429A2010-05-19
CN102426695A2012-04-25
CN106940897A2017-07-11
US20190369742A12019-12-05
US20130335445A12013-12-19
Attorney, Agent or Firm:
HANGZHOU QIUSHI PATENT OFFICE CO., LTD. (CN)
Download PDF: