Research On Deformation Technology Of 3D Models

Three dimensional geometric modeling is an important technology for the construction of personalized digital products,which is widely used in fields like film and animation,industrial design and biomedicine.3D model deformation is to transform the original model while preserving the connectivity information between vertices,thus creating a new 3D geometric model.It is an important step of the 3D geometric modeling.Based on the data of the existing 3D models,this thesis studies the deformation algorithm which is based on the shape characteristics of 3d models.The main research contents and innovations are summarized as follows:(1)This thesis studies the relationship between image shape features of 2D and 3D model,and proposes a 3D model deformation algorithm.Firstly,the shape features of 2D image and 3D model are analyzed,and the feature lines of image and model are obtained.Secondly,according to the image and model feature lines,the association relation between image and model shape feature is given by using the similar triangle relation,and the control constraint conditions are defined to drive the model deformation.According to the shape feature correlation between the image and the model after deformation,a method to evaluate the quality of deformation results by comparing the shape correlation between the image and the model after deformation was proposed.The experimental results show that the 3D model deformation algorithm guided by 2D image shape features proposed in this paper is automatic and efficient,which makes the modeling process more personalized while creating the 3D model.(2)To solve the problem that the rigid mesh deformation algorithm preserves the local characteristics but ignores the global ones,this thesis proposes an improved rigid mesh deformation algorithm which takes both the local and the global into account.The centroid difference function is constructed by using the change of the centroid of the model during the deformation process,and it is added into the original energy function to control the mesh deformation effect,so as to maintain the local rigid transformation of vertex elements and ensure the overall structural stability of the deformation model.According to the experimental results,the improved rigid mesh deformation algorithm proposed in this thesis can achieve better deformation effect with little change in time complexity.(3)The shape features and their correlation of multiple 3D models are studied,and a model deformation algorithm based on the correlation among the shape features of multiple models is proposed.First,the 3D model registration algorithm is studied.Multiple input models are registered in the same coordinate system to establish the corresponding relationship between model vertices and extract the feature point set of the original model as the shape feature.Second,based on the quasi-conformal mapping,the 3D model is parameterized to the 2D plane.Then,according to the correlation of shape features among multiple models,the control constraints of deformation are established to drive the two-dimensional plane deformation based on quasi-conformal iteration.Eventually,the model recovery algorithm is studied to realize the transformation of the model from two-dimensional plane to three-dimensional space under the premise of maintaining the shape characteristics.In the experiment,a qualitative evaluation method of deformation algorithm based on model shape feature is proposed.For the models represented by 3D face models with rich details and insignificant skeletons,the experiment proves that this deformation algorithm can achieve better deformation effect compared with the deformation algorithm based on differentiation.(4)The shape characteristics and correlation of 3D model components are studied,and a 3D model deformation algorithm combining global and local is proposed.First,a 3D model component segmentation and classification algorithm based on deep learning framework is proposed.Second,the shape characteristics of the segmented parts and similar parts in the model database are compared to study the correlation between them.The 3D model deformation algorithm based on image shape characteristics proposed in this paper is used to define control constraints to realize the deformation of model parts.Finally,spline suture technology is studied to realize 3d model reconstruction.The experimental results show that this algorithm is suitable for various types of model reconstruction and has good deformation effect and high timeliness.