Font Size: a A A

Research On Craniofacial Reconstruction Based On Statistical Deformable Model

Posted on:2010-09-28Degree:MasterType:Thesis
Country:ChinaCandidate:X Z OuFull Text:PDF
GTID:2178360272993926Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
In this paper, we present an approach to reconstruct face from skull based on 3D statistical models. After three-dimensional reconstruction from CT image series, features points are selected and then point distribution deformable models are established. Specifically, the statistical models are first partitioned into known and unknown parts, and then the unknown face is estimated by the prior knowledge constructed by the statistical model of the skull. Finally, a reconstruction system is designed and implemented. The main contents of this paper are listed as follows:1) Preprocessing of Skull and Face from CT ScansSkull and facial data from living subjects is acquired by CT-scanning. Some image preprocessing algorithms are used to remove noise and enhance the region of interesting. The marching Cubes algorithm is used to reconstruct the models of skulls and faces.2) Selection of Feature PointsAn interactive system is presented to extract feature points. We also proposed correspondence-matching algorithm based on an algorithm called Relative Angle Context Distribution. Through the comparison of the Relative Angle Context Distributions of the feature points among different models, the corresponding points are selected systematically.3) Establishment of the joint statistical model of skull and faceStatistical models of the skull and the face are built using Principal Component Analysis. A new sample can be calculated by a geometrically averaged facial template, with the addition of a correlation-ranked set of modes of principal variations.4) Reconstruction of face from the joint statistical model and the skullAs the prior knowledge of relationship between the face and the skull has been incorporated to the joint statistical model, the missing data of the face can be recovered from the model and the skull. The accuracy and efficiency of the recovery algorithm have been tested and analyzed comprehensively.5) Implementation of a Facial Reconstruction System (FRS)A facial reconstruction system called FRS is designed and implemented which including 3D reconstruction, landmark detection and registration etc. The system combine statistics with computer science, which made reconstruction be more reliability.
Keywords/Search Tags:Craniofacial Reconstruction, Statistical Model, Feature Point Extraction, 3D Reconstruction
PDF Full Text Request
Related items