Research On Rapid Skull Retrieval In Craniofacial Restoration

In the template-based craniofacial restorative method,the initial stage of restoration requires selecting the most similar reference skull in the database and then calculating the deformation between the skull to be restored and the reference skull,and finally applying the deformation to the skin of the reference skull,the restoration of the skull to be restored is completed.The choice of the reference skull is an application of 3D model retrieval.Due to the subtle differences between the skulls,the classic 3D model retrieval algorithm is not applicable.This article focuses on how to conduct fast and accurate 3D similar skull retrieval.The main research contents include:First,skull data preprocessing.Introduced the complete flow of living skull data from scanning to storage.First,three-dimensional reconstruction and denoising were performed on the CT skull data originally acquired to form a single layer of skull data.Second,Introducing the Frankfurt coordinate in craniofacial morphology to normalize skull data;Finally,the skull database was designed and constructed,and for the incomplete skull retrieval,skull partition storage was added.Second,a three-dimensional skull retrieval algorithm based on multi-scale projection is proposed.Firstly,the orthogonal projection and depth projection of the three-dimensional skull are obtained,and the skull profile angles,distances and profile features with spatial information are further obtained.Second,through the sample library supervised learning,the weighting coefficients corresponding to the characteristics of the different description capabilities of the skull shape are determined.Finally,based on ICP+TPS registration,the errors of the skull were searched.Experiments show that the skull features extracted by multi-scale projection can effectively characterize the skull,and the retrieval accuracy of skulls with different feature weighting coefficients is higher.Third,Partition-based incomplete skull retrieval.For the incomplete skull,this article introduces the idea of partitioning to search for the incomplete skull so that the intact reference skull can be retrieved for the incomplete skull.Experiments show that the algorithm can be used accurately in the selection of the most similar skull in the early stage of craniofacial reconstruction while guaranteeing retrieval efficiency.Based on the above research,this paper designs and implements a rapid skull retrieval prototype system.The system functions include skull database creation,skull feature extraction,most similar skull retrieval,and retrieval effect evaluation.Experiments show that the system designed in this paper can retrieve the most similar skull in the database for template-based craniofacial reconstruction.