Automatic labeling method for three-dimensional model component categories

Abstract

The invention discloses an automatic labeling method for three-dimensional model component categories. The method includes the step of a CRF (conditional random field) labeling model training process, namely, training and learning to obtain a CRF labeling model for segmenting and labeling an unknown three-dimensional model according to a three-dimensional model labeling training set: firstly, dividing the three-dimensional model labeling training set into an instance set and a validation set; then, preprocessing the three-dimensional model labeling training set in the instance set so as to train to obtain unitary items and binary items of the CRF labeling model by classification; and finally, using a preprocessing result of the validation set for parameter search to obtain parameters of the CRF labeling model so that learning of the CRF labeling model is completed. During labeling of the target three-dimensional model, the CRF labeling model obtained from a learning process of the three-dimensional model labeling training set is used for segmenting and labeling, so that category labeling of target three-dimensional model components is achieved.

Description

technical field [0001] The invention relates to a processing method for shape analysis, which belongs to the technical field of computer graphics, in particular to an automatic labeling method for three-dimensional model component categories. Background technique [0002] Segmenting a 3D model into meaningful components is the basis for shape understanding and high-level geometric processing. Further, identifying and obtaining labeling issues that describe the component categories of 3D models is also a problem in geometric modeling, 3D model animation and texture, etc. The key to many tasks in the field, for example, in the application of human mesh texture synthesis, it is necessary to identify the part of the mesh with "arm" texture, or the part with "leg" texture, etc.; in addition, some do not directly require segmentation labeling Applications such as 3D shape matching or retrieval can also benefit from component and annotation class information. [0003] Although a l...

AI Technical Summary

Problems solved by technology

[0003] Although a lot of work has been carried out on automatic image annotation, such as literature 1: Bao Hong, Xu Guangmei, Feng Songhe, Xu De. Research progress in automatic image annotation technology. Computer Science, 2011,38(7):35-40. However, the 3D model Most of the work in this field is only researched on the overall labeling of 3D models, such as literature 2: Tian Feng, Shen Xukun, Liu Xianmei, Zhou Kai, Du Ruishan. A method for automatic semantic labeling of 3D models based on weak labels, Journal of System Simulation, 2012, 24 (9): 1873-1876, 1881, but does not involve the automatic category labeling of 3D model components; in addition, the model segmentation that is inseparable from the 3D model component labeling, such as literature 3: Chen X., Golovinskiy A., Funkhouser T.A Benchmark for 3D Mesh Segmentation. ACM Transactions on Graphics, 2009, 28(3). The above is still an open research problem. Most model segmentation methods use simple and interpretable geometric algorithms, but are limited by general rules (eg, Concave shape, architectural topology, appropriate shape primitives) or a single feature (e.g., shape diameter, curvature tensor, geodesic distance) to divide the input mesh, cannot be applied to different types of 3D models, and it is even more difficult to achieve the same type of model Consistent segmentation of categories, so it is difficult to meet the category labeling requirements of component parts; recently, literature 4: Golovinskiy A., Funkhouser T.Consistent segmentation of 3D models.Computers and Graphics(Shape Modeling International09)2009,33(3):262- 269., Document 5: Xu Kai. Semantic-driven 3D shape analysis and modeling. [D] Graduate School of National University of Defense Technology. 2011. Considering that the 3D model of the same object contains more semantic information than a single model, Therefore, a joint segmentation method is proposed to analyze the same model set to obtain consistent segmentation of multiple models. However, this method does not consider the automatic labeling of unknown 3D models, and cannot automatically obtain the category information of the components of the 3D model; literature 6: Kalogerakis E., Hertzmann A., Singh K.. Learning 3D mesh segmentation and labeling. ACM Transactions on Graphics, 2010, 29(4) Article No.102. First proposed a learning method for model segmentation and labeling, they By learning the manually segmented and labeled model set, the model part labeling problem is expressed as a conditional random field optimization problem, so as to realize the unknown The segmentation and labeling of the model, however, the learning process of the method is time-consuming, and the method disclosed in the present invention further eliminates redundant samples through the symmetry detection of the three-dimensional model, thereby reducing the training time of the method

Images