Font Size: a A A

Research Of 3D Craniofacial Reconstruction Based On Partition Reconstruction

Posted on:2017-11-15Degree:MasterType:Thesis
Country:ChinaCandidate:Z Z LiFull Text:PDF
GTID:2348330512464264Subject:Communication and Information System
Abstract/Summary:PDF Full Text Request
Craniofacial reconstruction is a technology that is based on the intrinsic relationship between skull and face to reconstruct the unknown skull face. To study this technology deeply has vital significance for archaeology and medical plastic surgery and so on. This thesis focus on small sample problems that occur in the computer-aided craniofacial restoration process and issues lacking details of the reconstructed face, and it will propose the improved method based on partition theory to reconstruct the face. Initially, the craniofacial model will be divided into some segments by this method, such as the nose, the eyes and so on, subsequently, reconstruct these segments independently, finally splice craniofacial partition and get the reconstructed face.The main works are listed as follows:1. Use a method based on sphere partition histogram descriptor to calibrate the feature points, find these points that are similar to descriptor template according to this method, then get center points of these facial feature points. Furthermore, this method has strong positioning capability, but it is not sensitive to noise and model resolution, etc.2. Propose a partition method based on spectral clustering. This method will map the skull and facial points to the spectrum space, and then realize the division of the skull and face by partitioning spectral space area to realize the craniofacial partition. Moreover, experiments show this method makes the partition result more accurate.3. Propose a partial least squares regression recovery method based on spectral clustering. The various partitions of the skull, which will be restored, are treated as the known conditions. According to the locally related model of independent variable (vertex of the skull) and dependent variable (vertex of facial skin), each partition is obtained. The whole facial skin is recovered by stitching algorithm based on adjusting the boundary curvature.4. A skull restored prototype system is designed and realized. This system will realize three modules functions, which include the establishment of a database, division and recovery about skull and all these will be pointed out at the end of the article and all methods mentioned above will be verified.
Keywords/Search Tags:Craniofacial Recovery, Craniofacial Partition, Spectral Clustering, Partial Least Squares Regression
PDF Full Text Request
Related items