We propose a novel framework for 3D facial similarity measurement and its application in facial organization. The construction of the framework is based on Kendall shape space theory. Kendall shape space is a quotient space that is constructed by shape features. In Kendall shape space, the shape features can be measured and is robust to similarity transformations. In our framework, a 3D face is represented by the facial feature landmarks model (FFLM), which can be regarded as the facial shape features. We utilize the geodesic in Kendall shape space to represent the FFLM similarity measurement, which can be regarded as the 3D facial similarity measurement. The FFLM similarity measurement is robust to facial expressions, head poses, and partial facial data. In our experiments, we compute the distance between different FFLMs in two public facial databases: FRGC2.0 and BosphorusDB. On average, we achieve a rank-one facial recognition rate of 98%. Based on the similarity results, we propose a method to construct the facial organization. The facial organization is a hierarchical structure that is achieved from the facial clustering by FFLM similarity measurement. Based on the facial organization, the performance of face searching in a large facial database can be improved obviously (about 400% improvement in experiments).
Craniofacial reconstruction is to estimate a person’s face model from the skull. It can be applied in many fields such as forensic medicine, archaeology, and face animation. Craniofacial reconstruction is based on the relationship between the skull and the face to reconstruct the facial appearance from the skull. However, the craniofacial structure is very complex and the relationship is not the same in different craniofacial regions. To better represent the shape changes of the skull and face and make better use of the correlation between different local regions, a new craniofacial reconstruction method based on region fusion strategy is proposed in this paper. This method has the flexibility of finding the nonlinear relationship between skull and face variables and is easy to solve. Firstly, the skull and face are divided into five corresponding local regions; secondly, the five regions of skull and face are mapped to low‐dimensional latent space using Gaussian process latent variable model (GP‐LVM), and the nonlinear features between skull and face are extracted; then, least square support vector regression (LSSVR) model is trained in latent space to establish the mapping relationship between skull region and face region; finally, perform regional fusion to achieve overall reconstruction. For the unknown skull, first divide the region, then project it into the latent space of the skull region, then use the trained LSSVR model to reconstruct the face of the corresponding region, and finally perform regional fusion to realize the face reconstruction of the unknown skull. The experimental results show that the method is effective. Compared with other regression methods, our method is optimal. In addition, we add attributes such as age and body mass index (BMI) to the mappings to achieve face reconstruction with different attributes.
14 January 2026 in Expert Systems with Applications