Volumetric deformable face models for cranio-facial reconstruction

Abstract

In forensic cranio-facial reconstruction, facial features of an unknown individual are estimated from an unidentified skull, based on a mixture of experimentally obtained guidelines on the relationship between soft tissues and the underlying skeleton. In this paper, we investigate the possibility of using full 3-D cross-sectional CT images for establishing a reference database of densely sampled distances between the external surfaces of the skull and skin for automated craniofacial reconstruction. For each CT image in the reference database, the hard tissue (skull) and extra-cranial soft tissue (skin) volumes are segmented and transformed into signed Distance Transform (sDT) maps. A simplified procedure for cranio-facial reconstruction was implemented, by warping all reference skull sDT maps to the target skull sDT. These warps are subsequently applied to the reference skin sDT maps and the zero level set of their arithmetic average is defined as the reconstructed target skin surface. Initial results are shown to proof the validity of the concept, but further refinement of the procedures involved and a qualitative validation are required.

Bayesian estimation of optimal craniofacial reconstructions

“…This implicit surface representation encodes for every 3D point an approximated shortest Euclidian distance perpendicular to the skull surface, zero on the surface, positive outside and negative inside. This representation is very similar to the one used in [31][32][33]. The advantage of such an implicit representation is that information about the skull is known for every point on and around the skull.…”

“…Model bias is the leaking through of unwanted specific facial feature details, originating from the CFT, into the reconstruction. This model bias can be reduced by using multiple reference heads, as in [28][29][30][31][32][33]. Second, the applied generic deformations are not facespecific, they are just smooth.…”

“…For representations containing also skull surface information into the CFM, more dense representations are used like Crestlines [13] and feature points [14,15,19]. Similar to the method reported in this paper, an alternative implicit TSR is used in [31][32][33], where the skull surface is represented using a signed Distance Transform, representing for each voxel in the CT image the Euclidean distance to the closest point on the skull and head surface, respectively, distances being positive (negative) for voxels inside (outside) the skull/head.…”

Segmentation of head bones in 3-D CT images from an example

“…Devising reliable and fully automatic segmentation methods for head bones in 3-D CT images is still an open issue, and few methods in the literature are addressing this problem [1,2]. One of the key points making this task difficult is related to the nature of CT images, which are subject to noise and to severe streaking artifacts due to metal objects such as dental fillings.…”

A Three-Dimensional Computerized Facial Reconstruction Using Non-Linear Registration of A Reference Head

“…Thus, it is also prominent to use them for facial reconstruction as well. Generally, these techniques are based on deformation approaches of a reference head to unknown skull based on crest lines [12], disc fields [7,13], feature points [14], Euclidean distance maps [15], or semi-landmarks [16]. The reference head is often selected based on similarities in morphological characteristics (i.e.…”