A 3D Model Generation Method Based on Isomorphic Model Representation
A 3D model and structure representation technology, applied in neural learning methods, biological neural network models, neural architectures, etc., can solve problems such as heavy user burden, time-consuming, and small difference in modeling results
Active Publication Date: 2020-06-05
NANJING UNIV
View PDF2 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
3D modeling often requires user interaction and consumes a lot of time, and many applications have certain requirements for the number of 3D models. Therefore, how to quickly and effectively build large-scale 3D models is an important research direction for 3D modeling.
This method can generate models with large differences, but during the modeling process, user interaction is very frequent, and it takes a lot of time to build a model, so this method is also difficult to generate a large number of 3D models
All in all, there are major defects in the existing technology: first, the modeling process takes a long time, and it is difficult to build a certain number of model sets; second, the difference in modeling results is small, and the generated result model set is relatively monotonous; third, it needs more Many interactions, heavy user burden
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment
[0100] In this embodiment, 13 chair 3D models are used as the system input model library, and the 13 chairs are unified isomorphic representations, and their respective representation vectors are calculated, and the representation vectors are two-dimensionally encoded through a three-layer fully connected neural network , and sample in the two-dimensional coding space, check the validity, visualize the effective space and render it, and generate a new model according to the user's click operation.
[0101] The specific implementation process is as follows:
[0102] In step 1, the user imports the model set, and then executes the generation of the unified structure diagram in step 11. First extract the structure graph of each model, such as figure 2 shown in the second column from left to right. After that, the unified structure diagram is extracted, and the extracted unified structure diagram is shown in figure 2 The third column from left to right. After extracting the ...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses an isomorphic model representation-based three-dimensional model generation method. The method comprises the steps of establishing a unified structure representation of a modelset for corresponding part relationships of a model set; in a sub-graph coding way, establishing an isomorphic structure representation of each model; by adopting a bounding box and a generalized cylinder, establishing a part representation of each model; according to the structure representations and the part representations of the models, establishing a unified representation of the models; training a neural network-based autocoder, and establishing a mapping relationship between an isomorphic representation space and a two-dimensional numerical space; performing sampling on two-dimensionalnumerical values, and performing decoding by utilizing the autocoder to obtain an isomorphic model representation; according to the isomorphic model representation obtained by decoding, re-building athree-dimensional model, and judging the validity of the re-built model; according to the validity of sampled data, estimating distribution of effective spaces, and performing visualization; and according to effective two-dimensional data selected by a user, performing decoding to obtain the model representation, and re-building a new three-dimensional model.
Description
technical field [0001] The invention belongs to the technical field of computer graphics, in particular to a three-dimensional model generation method based on isomorphic model representation. Background technique [0002] 3D modeling is an important task of computer graphics and the basis for other follow-up research and applications. 3D modeling often requires user interaction and consumes a lot of time, and many applications have certain requirements for the number of 3D models. Therefore, how to quickly and effectively build large-scale 3D models is an important research direction of 3D modeling. [0003] In fact, the direction of 3D modeling has produced many related technologies and methods. The traditional CAD modeling method requires the modeler to start from the underlying vertices and patches, and perform a large number of interactions, such as inputting vertex coordinates, moving patches, etc. to complete the modeling process. [0004] Due to the heavy interacti...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06T17/00G06N3/04G06N3/08
Inventor 孙正兴武蕴杰宋有成宋沫飞
Owner NANJING UNIV