Research On Similarity Measure Of Craniofacial Shape Based On Feature Point

Compared with traditional craniofacial rehabilitation methods,computer-assisted craniofacial restoration has obvious advantages,and its recovery results are closer to the real surface model.The craniofacial similarity measurement method can not only evaluate the recovery results,but also evaluate the restoration methods and improve the existing restoration methods.Therefore,the craniofacial similarity measurement has gradually become a hot topic in craniofacial morphology research.In this paper,the three-dimensional craniofacial model as the research object,focusing on the study of craniofacial radial feature extraction methods,and adaptive neighborhoods used in similarity measurement methods,designed to solve the automatic measurement of craniofacial model similarity.The main work is summarized as follows:1.The three-dimensional craniofacial model was reconstructed and standardized.The part includes craniofacial extraction based on CT data,craniofacial reconstruction,unified Frankfurt coordinates and scale normal.Reconstructed a single-layered skull and real-facial model from CT data,the real-facial model and the reconstruction coordinate system and scale were normalized,which provided a good data foundation for follow-up study.2.The traditional method of facial similarity is easily affected by the subjective consciousness and the limitations for semi-automatic extract the key features is inaccurate.To solve this problem,an automatic craniofacial similarity measurement algorithm based on radial feature points was proposed.Integrating the experience knowledge of face recognition,firstly,extract the contour lines with different intensity levels begun from the tip of nose and ended at the boundary of the face model,extract the feature points on the contour lines by the geodesic distance.Shape factor measures the feature points,and the similarity between appearance models is measured by correlation coefficients.Experiments show that the method can automatically extract feature points and measure the similarity of the three-dimensional appearance model,and the measurement results are consistent with the subjective evaluation.3.Aiming at the problem that the real face mode with expression feature,while the reconstructed one with nothing,3D Face Similarity Measurement Based on Adaptive Neighborhood Description was proposed.This paper analyzes the changes of feature points on facial contours of different expressions,uses adaptive neighborhood constraints to select neighborhood points that are not associated with facial expression features,and participates in constructing geometric feature descriptors of feature points.The degree of similarity between the models to be measured is measured on the basis of the geodesic distance.The experimental results show that the method can effectively measure the similarity of the three-dimensional face model,and has good robustness to the face model with expression features.