Craniofacial Statistical Restoration Based On Curves

Craniofacial reconstruction is a research that reconstructs the corresponding facial features based on the human skull,and has been widely used in many fields such as public security forensics,archeology,and medical plastic surgery.In order to reduce the dimension of the sample data in the model and improve the accuracy of reconstruction,this thesis extracts two types of curves that are evenly distributed on the human face from the nasal point on the three-dimensional craniofacial model.The skull data and the two types of curves are used as training sample data,and the statistical models those combine the two types of curves and the skull data are established respectively.The prior knowledge obtained by the statistical model and the skull data to be reconstructed are used to estimate the facial data.The main innovation and research work of this thesis are as follows:1.Craniofacial reconstruction based on principal component analysis(PCA)Based on the PCA-based craniofacial statistical reconstruction model,the skull and face are used as training samples to establish a statistical model composed of skull data and facial data.Input the known face data to obtain the recovered skull data;enter the known skull data to obtain the recovered face data.The reconstruction results are compared with the original skull and the original face respectively.Euclidean distance is used to calculate the error between the reconstruction result and the original skull and original face,respectively,to measure the similarity between the reconstruction result and the original model.2.Craniofacial statistical reconstruction based on radial curvesUsing the skull data and the radial curves extracted from the corresponding face as training sample data,establish a statistical model of the skull and radial curves,using the prior knowledge obtained by the statistical model plus the skull data to be reconstructed to reconstruct the radial curves of the face corresponding to the unknown skull,and finally,the face of the skull to be reconstructed is estimated according to the face statistical model.Compare the reconstructed face obtained under this method with the corresponding original face,use the Euclidean distance to calculate the error between the reconstructed face and the original face,and compare this error with the previously reconstructed PCA craniofacial statistics.Comparing the appearance errors,the results show that the craniofacial statistical reconstruction based on the radial curve has a smaller error,which means that the effect of craniofacial statistical reconstruction based on radial curves is better than that of PCA craniofacial statistical reconstruction.3.Craniofacial statistical reconstruction based on geodesicsUsing the skull data and the radial curves extracted from the corresponding face astraining sample data,establish a statistical model of the skull and geodesics,and use the prior knowledge obtained by the statistical model plus the skull data to be reconstructed to reconstruct the radial curves of the face corresponding to the unknown skull,and finally,a new face is estimated according to the face statistical model.Compare the reconstructed face obtained under this method with the corresponding original face,use the Euclidean distance to calculate the error between the reconstructed face and the original face,and compare this error with the previously calculated error under the PCA craniofacial statistical reconstruction.The results show that the craniofacial statistical reconstruction based on geodesics has smaller error in reconstruction results,which means that the effect of craniofacial statistical reconstruction based on geodesics is better than that of PCA craniofacial statistical reconstruction.Compared with the PCA-based craniofacial reconstruction method,the curves-based craniofacial statistical reconstruction method proposed in this paper reduce the dimensions of the sample data in the model,improve the accuracy and speed of craniofacial reconstruction,and the effect of craniofacial reconstruction is further improved.