Research On Semantic Feature Extraction And Deformation Algorithm For Geometric Human Body

In recent years,with the appearance of low-price sensors like Kinect,acquiring 3D object has become more and more fast and convenient.The reconstructed 3D human bodies can be used for many applications,such as: personalized furniture customization,virtual try-on,etc.These applications cannot be realized without the help of 3D human body semantic features(such as,anthropometric data,feature points)extraction and deformation algorithms.Therefore,it is a meaningful work to design algorithms for automatic semantic feature extraction and deformation of 3D human body.In the future,there will be more and more realistic 3D human bodies reconstructed by 3D scanners and developing mature semantic feature extraction and deformation algorithms will be a valuable job.At present,3D human body semantic feature extraction methods include automatic segmentation,feature point extraction and so on.In this paper,we will design an efficient 3D human body semantic feature extraction algorithm base on both methods mentioned above.On the other hand,methods of mesh deformation include skinning methods,data driven methods,and physical methods.And we will design a 3D human model deformation algorithm that deforms the human body to match the product according to anthropometric data and product data.In this paper,we will discuss how to divide a 3D human body mesh reconstructed through 3D human body modelling technology into several parts according to human semantic,and how to extract the anatomical landmarks as well as compute anthropometric data from the human body.In the meanwhile,we will also discuss how to deform a 3D human model to a proper pose to match the product(such as: sitting a human body model on a wheelchair model)so that we can make a demo for the personalized product.To achieve this,we adopted an automatic human body segmentation algorithm based on skeletal curves for segmentation,then we extract anatomical landmarks through heat kernel signature and curvatures,finally we deform the 3D human body with the guidance of human semantic constraints thorough a blend skinning method based on optimized rotation centre.The 3D human body segmentation and deformation algorithm for semantic constraints of product design can well assist the design of personalized product,making custom product design more flexible so that we can have more personalized products and facilitate people's live.