Real-time sparse editing method and system based on large-scale grid model

Abstract

The invention discloses a real-time sparse editing method based on a large-scale grid model. The real-time sparse editing method comprises the following steps: constructing a model library consistingof models with the same grid topology; extracting an original feature vector of the model, converting the original feature vector into a first feature vector capable of encoding large-scale deformation of the model, and forming a first feature vector set of the model library; performing global sparse optimization on the first feature vector set, and extracting a second feature vector set; determining a control vertex position on the target model; taking the feature vector of the target model as a linear combination vector of the second feature vector set, and performing sparse optimization onthe linear combination vector to obtain an editing feature vector; and obtaining a deformation model of the target model according to the control vertex position and the editing feature vector. According to the real-time sparse editing method, a small number of deformation features can be automatically selected through a sparse mixing method by means of prior information of a model base, and a more real, natural and reliable editing result is generated in real time.

Description

technical field [0001] The invention relates to the technical fields of digital media and computer graphics, in particular to a real-time sparse editing method based on a large-scale grid model. Background technique [0002] The model editing technology can model and generate a 3D model that meets the editing conditions according to the user's editing. The key to the model deformation technology is to generate a natural and reasonable 3D model on the basis of reducing the user's workload. The instance-based mesh deformation method is a powerful tool for model editing. However, existing techniques often combine all instance deformation modes, which can lead to overfitting, i.e., using an overly complex model to account for user-specified deformations. This will lead to impossible or unstable deformation results, including some global changes outside the region of interest. [0003] Many works in the traditional method edit the 3D model directly without using the prior infor...

AI Technical Summary

Problems solved by technology

However, there are few related works using the model library for analysis. At present, the most relevant work, MeshIk and RIMD, uses deformation gradients as features. Although the latter can handle large-scale deformations, its representation cannot display derivatives, and requires all Deformation features are globally mixed, which often leads to low computational efficiency, and some unnecessary deformations in the final results

Therefore, it is still vacant to efficiently generate more robust results through reasonable selection of deformation features in the model library.

Images