3D craniofacial similarity calculation and craniofacial relationships analysis based on spectral analysis method

The field of 3D craniofacial similarity calculation and craniofacial relationships(the relationships between the human skull and face) analysis is a challenging and meaningful task in archaeology, forensic science, and anthropology. Although anthropologists have been involved in the application and illustration of craniofacial relationships in their research, due to the complexity of the 3D skull (with multiple holes), and the facial expression changes of 3D human faces, it is difficult to accurately perform 3D craniofacial similarity calculations and theoretical validation of craniofacial relationships are difficult. To address this challenge, we propose a data-driven framework that constructs a shape feature space based on spectral analysis to describe the intrinsic structure of 3D skulls and faces. Our framework includes a shape analysis method to measure 3D craniofacial similarity and a statistical method using canonical correlation analysis to comprehensively describe the craniofacial relationships from global statistical features and individual geometric features. Based on an Asian craniofacial database, we demonstrate the effectiveness of our framework through validation results and skull identification. Most importantly, we provide two craniofacial relationships rules through theoretical validation and numerical results: R1-the human skull has a strong correlation with the face; R2-the similarity change trend of the skull is generally consistent with the corresponding face similarity. These craniofacial relationships rules can be applied to general craniofacial analysis tasks and scenarios, providing a solid theoretical basis for relevant researchers. Our research represents a significant contribution to the field of 3D craniofacial similarity calculation and craniofacial relationships analysis.