Automatic skull-face overlay and mandible articulation in data science by AIRS-Genetic algorithm

Abstract

Craniofacial superimposition (CFS) is an innovative method of identifying the missing skull of a human and has received serious attention in forensic research. Optimal overlaying of skull face is one of the fundamental issues found on forensic technology. CFS methodology generally initiates with the marking the accurate landmarking of skull and face extracted from the scanning device which is termed as ground truth data. These extracted features of skull and face will be in crisp image as 3D and 2D. Consequent of this advanced technology, CFS has upgraded to automatic level to investigate different cases. Regardless, the forensic expert invests more for investigation and it consumes large time for completing it. Most of the review study proofs that the computer-aided methods can improve the efficiency of the CFS technique can improve much better than other techniques. The introduction of computerized 3D-2D CFS in forensic research proves to be a novel one and define as most advance CFS technique in research today. This methodology can automate the operation of CFS by reducing the error on over-layered model. Due to the involvement of machine learning process in forensic field, executing the CFS problem is much easier. Coevolutionary algorithms (CEAs) are nature-inspired and very optimistic in solving complex problems by measuring the fitness of the problem. In this research work, automatic CFS is proposed for skull-face overlay and mandible articulation. Conventionally, a fuzzy set is proposed for extracting the features of skull especially the landmarking. In this work the Artificial Immune Recognition System (AIRS) model is implemented for distance calculated for crisp point of the extracted image. Due to the increase of computation time for crisp landmarks, an efficient algorithm is needed. With the point of decreasing the time of computation not by losing the accuracy the cooperative coevolutionary algorithm is proposed. Similar to the AIRS function, the Euclidean distance metrics are calculated between the AIRS set and crisp points. Hence the fitness is calculated form the algorithm for better layout of landmarking for better skull face overlying. Ultimately, the outcome of the progression is compared with the existing methodologies for measuring the level of effectiveness respectively.