Data-driven human body morphing

Show simple item record

dc.contributor.advisor Ergun , Akleman en_US
dc.contributor.committeeMember Bortfeld , Heather en_US
dc.creator Zhang , Xiao en_US 2005 -11 -01T15 :48 :36Z 2014 -02 -19T19 :22 :42Z 2005 -11 -01T15 :48 :36Z 2014 -02 -19T19 :22 :42Z 2005 -08 en_US 2005 -11 -01T15 :48 :36Z
dc.identifier.uri http : / /hdl .handle .net /1969 .1 /2655
dc.description.abstract This thesis presents an efficient and biologically informed 3D human body morphing technique through data -driven alteration of standardized 3D models . The anthropometric data is derived from a large empirical database and processed using principal component analysis (PCA ) . Although techniques using PCA are relatively commonplace in computer graphics , they are mainly used for scientific visualizations and animation . Here we focus on uncovering the underlying mathematical structure of anthropometric data and using it to build an intuitive interface that allows the interactive manipulation of body shape within the normal range of human variation . We achieve weight /gender based body morphing by using PCA . First we calculate the principal vector space of the original data . The data then are transformed into a new orthogonal multidimensional space . Next , we reduce the dimension of the data by only keeping the components of the most significant principal vectors . We then fit a curve through the original data points and are able to generate a new human body shape by inversely transforming the data from principal vector space back to the original measuring data space . Finally , we sort the original data by the body weight , calculating males and females separately . This enables us to use weight and gender as two intuitive controls for body morphing . The Deformer program is implemented using the programming language C++ with OPENGL and FLTK API . 3D and human body models are created using Alias MayaTm . en_US
dc.format.extent 2646279 bytes
dc.format.medium electronic en_US
dc.format.mimetype application /pdf
dc.language.iso en _US en_US
dc.publisher Texas A &M University en_US
dc.subject Data -driven en_US
dc.title Data -driven human body morphing en_US
dc.type Book en
dc.type.genre Electronic Thesis en_US
dc.type.material text en_US
dc.format.digitalOrigin born digital en_US


Data-driven human body morphing. Available electronically from http : / /hdl .handle .net /1969 .1 /2655 .

Files in this item

Files Size Format View

There are no files associated with this item.

This item appears in the following Collection(s)

Show simple item record

Search DSpace

Advanced Search