Method and system for generating three-dimensional garment model

Patent metadata

Field	Value
Publication number	US-9940749-B2
Application number	US-201615062926-A
Country	US
Kind code	B2
Filing date	Mar 7, 2016
Priority date	Jul 13, 2015
Publication date	Apr 10, 2018
Grant date	Apr 10, 2018

Abstract

The present invention provides a method and a system for generating a three-dimensional garment model, where garment component composition information and attribute information corresponding to each garment component are acquired by acquiring and processing RGBD data of a dressed human body, and then a three-dimensional garment component model corresponding to the attribute information of each garment component is selected in a three-dimensional garment component model library, that is, a three-dimensional garment model can be constructed rapidly and automatically only with RGBD data of a dressed human body, and human interactions are not necessary during the process of construction, thus the efficiency of a three-dimensional garment modeling is improved, and it has significant meaning for the development of computer-aided design, three-dimensional garment modeling and virtual garment fitting technology.

First claim

What is claimed is:

1. A method for generating a three-dimensional garment model, comprising:
acquiring Red Green Blue and Depth (RGBD) data of a dressed human body;
constructing a garment information expression tree, wherein the information expression tree comprises a tree table containing garment component composition information and attribute information of each garment component;
constructing a garment image library, wherein the garment image library comprises garment images having a garment component pixel area identifier and a corresponding component composition information identifier, and wherein garment images in the garment image library further comprise Histogram of Oriented Gradient (HOG) feature and local binary pattern (LBP) feature identifiers in a pixel area of a garment component and an attribute information identifier of the corresponding component, wherein
the constructing the garment image library particularly comprises: according to the garment information expression tree, adding the garment component pixel area identifier and the corresponding component information identifier and HOG feature and LBP feature identifiers in the pixel area of the garment component and a corresponding attribute information identifier for garment images in the garment image library;

acquiring garment component composition information according to the RGBD data of the dressed human body, wherein
the acquiring garment component composition information according to the RGBD data of the dressed human body particularly comprises: determining a pixel area of each garment component according to the RGBD data of the dressed human body; acquiring component information corresponding to the pixel area in the garment image library, according to the pixel area of each garment component;

acquiring attribute information of each garment component based on the garment component composition information, wherein
the acquiring attribute information of each garment component based on the garment component composition information particularly comprises: extracting an HOG feature and an LBP feature in the pixel area of each garment component; acquiring attribute information of corresponding component in the garment image library, according to the HOG feature and the LBP feature in the pixel area of each garment component;

constructing a three-dimensional model library of the garment component, wherein the three-dimensional model library of the garment component comprises a three-dimensional garment component model with an attribute information identifier of the garment component, wherein
the constructing the three-dimensional model library of the garment component particularly comprises: according to the garment information expression tree, adding an attribute information identifier of the garment component for the three-dimensional garment component model in the three-dimensional model library of the garment component;

retrieving a three-dimensional garment component model corresponding to the attribute information of each garment component in a three-dimensional garment component model library; and
generating a three-dimensional garment model by assembling the three-dimensional garment component model.

2. The method according to claim 1 , wherein, after acquiring attribute information of each garment component, the method further comprises:
optimizing by inputting the attribute information of each garment component in a Bayesian network model, wherein the Bayesian network model is obtained by training garment images in the garment image library.

3. The method according to claim 1 , wherein, after generating the three-dimensional garment model by assembling the three-dimensional garment component model, the method further comprises
acquiring a three-dimensional posture and a garment point cloud of the human body according to the RGBD data of the dressed human body;
adjusting the three-dimensional garment model according to three-dimensional posture of the human body, so that the adjusted three-dimensional garment model can fit the garment point cloud.

4. A system for generating a three-dimensional garment model comprising a memory, a three-dimensional garment component model library, a garment component detector, a garment component attribute classifier, a garment information expression tree, a garment image library and a processor, wherein the memory is configured to store instructions, the processor is coupled with the memory and is configured to execute the instructions stored on the memory, and the processor is configured to:
acquire Red Green Blue and Depth (RGBD) data of a dressed human body;
wherein the garment component detector is configured to acquire garment component composition information according to the RGBD data of the dressed human body; and the garment component attribute classifier is configured to acquire attribute information of each garment component based on the garment component composition information;
wherein the processor is further configured to: retrieve a three-dimensional garment component model corresponding to the attribute information of each garment component in a three-dimensional garment component model library; and
generate a three-dimensional garment model by assembling the three-dimensional garment component model;
the garment information expression tree is a tree table containing garment component composition information and attribute information of each garment component;
the garment image library comprises garment images having a garment component pixel area identifier and a corresponding component information identifier, and Histogram of Oriented Gradient (HOG) feature and local binary pattern (LBP) feature identifiers in the pixel area of the garment component and a corresponding attribute information identifier, wherein, the garment component pixel area identifier and the corresponding component information identifier, and HOG feature and LBP feature identifiers in the pixel area of the garment component and the corresponding attribute information identifier are all added according to the garment information expression tree;
the garment component detector is specifically configured to determine a pixel area of each garment component according to the RGBD data of the dressed human body, and acquire component information corresponding to the pixel area in the garment image library, according to the pixel area of each garment component; wherein, the garment component detector is obtained by training garment images in the garment image library;
the garment component attribute classifier is specifically configured to extract an HOG feature and an LBP feature in the pixel area of each garment component, and acquire attribute information of corresponding garment component in the garment image library, according to the HOG feature and the LBP feature in the pixel area of each garment component; wherein, the garment component attribute classifier is obtained by training garment images in the garment image library;
the three-dimensional model library of the garment component comprises a three-dimensional garment component model with an attribute information identifier of the garment component, and the attribute information identifier of the garment component is added according to the garment information expression tree.

5. The system according to claim 4 , further comprises a Bayesian network model obtained by training garment images in the garment image library, and is configured to optimize attribute information of each garment component; and
the processor is further configured to acquire a three-dimensional posture and a garment point cloud of the human body according to the RGBD data of the dressed human body; adjust the three-dimensional garment model according to three-dimensio

Assignees

• Univ Beihang

Inventors

• Chen Xiaowu
• Zhou Bin
• Zhao Qinping
• Lu Feixiang
• Wang Lin
• Bi Lang

Classifications

• G06V10/764

  using classification, e.g. of video objects · CPC title

• G06F18/29

  Graphical models, e.g. Bayesian networks · CPC title

• G06F18/24155

  Bayesian classification · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06V10/56

  relating to colour · CPC title