Mikoláš Jurda

Testing photogrammetry-based techniques for three-dimensional surface documentation in forensic pathology

Three-dimensional surface technologies particularly close range photogrammetry and optical surface scanning have recently advanced into affordable, flexible and accurate techniques. Forensic postmortem investigation as performed on a daily basis, however, has not yet fully benefited from their potentials. In the present paper, we tested two approaches to 3D external body documentation - digital camera-based photogrammetry combined with commercial Agisoft PhotoScan(®) software and stereophotogrammetry-based Vectra H1(®), a portable handheld surface scanner. In order to conduct the study three human subjects were selected, a living person, a 25-year-old female, and two forensic cases admitted for postmortem examination at the Department of Forensic Medicine, Hradec Králové, Czech Republic (both 63-year-old males), one dead to traumatic, self-inflicted, injuries (suicide by hanging), the other diagnosed with the heart failure. All three cases were photographed in 360° manner with a Nikon 7000 digital camera and simultaneously documented with the handheld scanner. In addition to having recorded the pre-autopsy phase of the forensic cases, both techniques were employed in various stages of autopsy. The sets of collected digital images (approximately 100 per case) were further processed to generate point clouds and 3D meshes. Final 3D models (a pair per individual) were counted for numbers of points and polygons, then assessed visually and compared quantitatively using ICP alignment algorithm and a cloud point comparison technique based on closest point to point distances. Both techniques were proven to be easy to handle and equally laborious. While collecting the images at autopsy took around 20min, the post-processing was much more time-demanding and required up to 10h of computation time. Moreover, for the full-body scanning the post-processing of the handheld scanner required rather time-consuming manual image alignment. In all instances the applied approaches produced high-resolution photorealistic, real sized or easy to calibrate 3D surface models. Both methods equally failed when the scanned body surface was covered with body hair or reflective moist areas. Still, it can be concluded that single camera close range photogrammetry and optical surface scanning using Vectra H1 scanner represent relatively low-cost solutions which were shown to be beneficial for postmortem body documentation in forensic pathology.

Testing the reliability of software tools in sex and ancestry estimation in a multi-ancestral Brazilian sample

In the framework of forensic anthropology osteometric techniques are generally preferred over visual examinations due to a higher level of reproducibility and repeatability; qualities that are crucial within a legal context. The use of osteometric methods has been further reinforced by incorporating statistically-based algorithms and large reference samples in a variety of user-friendly software applications. However, the continued increase in admixture of human populations have made the use of osteometric methods for estimation of ancestry much more complex, which confounds one of major requirements of ancestry assessment - intra-population homogeneity. The present paper tests the accuracy of ancestry and sex assessment using four identification software tools, specifically FORDISC 2.0, FORDISC 3.1.293, COLIPR 1.5.2 and 3D-ID 1.0. Software accuracy was tested in a sample of 174 documented human crania of Brazilian origin composed of different ancestral groups (i.e., European Brazilians, Afro-Brazilians, and Japanese Brazilians and of admixed ancestry). The results show that regardless of the software algorithm employed and composition of the reference database, all methods were able to allocate approximately 50% of Brazilian specimens to an appropriate major reference group. Of the three ancestral groups, Afro-Brazilians were especially prone to misclassification. Japanese Brazilians, by contrast, were shown to be relatively easily recognizable as being of Asian descent but at the same time showed a strong affinity towards Hispanic crania, in particularly when the classification based on FDB was carried out in FORDISC. For crania of admixed origin all of the algorithms showed a considerable higher rate of inconsistency with a tendency for misclassification into Asian and American Hispanic groups. Sex assessments revealed an overall modest to poor reliability (60-71% of correctly classified specimens) using the tested software programs with unbalanced individual rates for males and females. The highest and atypically balanced rate of classification for sex assessment was provided by COLIPR software, which reached 78% of correctly assessed crania.

Using drone-mounted cameras for on-site body documentation: 3D mapping and active survey

Recent advances in unmanned aerial technology have substantially lowered the cost associated with aerial imagery. As a result, forensic practitioners are today presented with easy low-cost access to aerial photographs at remote locations. The present paper aims to explore boundaries in which the low-end drone technology can operate as professional crime scene equipment, and to test the prospects of aerial 3D modeling in the forensic context. The study was based on recent forensic cases of falls from height admitted for postmortem examinations. Three mock outdoor forensic scenes featuring a dummy, skeletal remains and artificial blood were constructed at an abandoned quarry and subsequently documented using a commercial DJI Phantom 2 drone equipped with a GoPro HERO 4 digital camera. In two of the experiments, the purpose was to conduct aerial and ground-view photography and to process the acquired images with a photogrammetry protocol (using Agisoft PhotoScan® 1.2.6) in order to generate 3D textured models. The third experiment tested the employment of drone-based video recordings in mapping scattered body parts. The results show that drone-based aerial photography is capable of producing high-quality images, which are appropriate for building accurate large-scale 3D models of a forensic scene. If, however, high-resolution top-down three-dimensional scene documentation featuring details on a corpse or other physical evidence is required, we recommend building a multi-resolution model by processing aerial and ground-view imagery separately. The video survey showed that using an overview recording for seeking out scattered body parts was efficient. In contrast, the less easy-to-spot evidence, such as bloodstains, was detected only after having been marked properly with crime scene equipment.

Sex and ancestry assessment of Brazilian crania using semi-automatic mesh processing tools

The present paper aims to test performances of semi-automatic tools for mesh-to-mesh processing while assessing sex and ancestry in documented human crania. The studied sample of 80 human crania, which originated in two documented Brazilian collections (São Paulo, Brazil) was digitized using photogrammetry and laser scanning. 3D cranial morphology was quantified by computing inter-mesh dissimilarity measures using in-house freeware FIDENTIS Analyst (www.fidentis.com). Numerical outputs were further processed using Discriminant Function Analysis and Canonical Variant Analysis in order to classify models into sex and ancestry groups. In addition, cranial morphology was described by a set of 37 landmarks, processed by a Procrustes analysis and confronted with the inter-mesh comparison. Patterns of sexual dimorphism and ancestral group-specific variation were interpreted using average meshes and further emphasized by employing advanced visualization graphics. The mesh-to-mesh processing was capable to detect shape differences related to sex and ancestry. The highest accuracy levels for sex determination were obtained for meshes representing the facial skeleton and the supraorbital region. For both, analysis correctly assigned 82.5% of the crania. Ancestry-related differences were manifested primarily in the global cranial features (observed accuracy rates reaching 63%). The advanced visualization tools provided a highly informative insight into sexual dimorphism and ancestry-related variation. While in the current state the technique cannot be considered suitable for being implemented into the everyday forensic practice, the extent of automatization proved to be perspective, especially for assessing skeletal features that cannot be properly quantified using discrete variables.

Applying 3D prints to reconstructing postmortem craniofacial features damaged by devastating head injuries

Postmortem facial identification is one of the most common techniques for establishing a deceased persons identity. In victims suffering from devastating cranial injuries, the feasibility of facial identification tasks can be compromised by damage to or disfigurement of the identifying cranial features. Although there are several reconstructive approaches, which help experts to restore the essence of persons physical appearance, thus enhancing the chances of recognition, only a few of them involve restoring the fractured cranial bones as the foundation for the reconstructed soft tissues. Here, we propose a technique based on replacement of heavily damaged hard tissues with generic prosthetics manufactured by 3D printing. Our approach does not require medical imaging technologies or other costly lab equipment. It is simple, affordable and relatively labor-efficient. The deceaseds reconstructed craniofacial features can be subsequently assessed, photographed, drawn or otherwise reproduced in order to help determine his or her identity. In addition, the imagery can be displayed, published or broadcasted in media without concerns of being overly graphic.

Introducing the FIDENTIS 3D Face Database

Abstract Face databases have assumed an important role in a variety of clinical and applied research domains. However, the number of datasets accessible to the scientific community is limited and the knowledge of their existence may be concealed from a wider range of specialists. In the present paper we introduce a sizeable dataset of 3D facial scans - FIDENTIS 3D Face Database (F3D-FD or FIDENTIS Database), which is accompanied by basic demographic and descriptive data. The database is structured according to recorded subjects, and comprises single-scan entries as well as a smaller number of multiscan entries. The multi-scan entries vary in the time passed between recording sessions and in the devices employed to collect the 3D data. The total number of 2476 individuals puts our database within the category of large-scale databases. The 3D scans are accessible through a web-based interface at www. fidentis.cz. A licensed version of the database is available to interested parties upon signing a license agreement. Because of its varied composition, and low target-specificity the database has capacity to be of great assistance for the worldwide research community.

VEAAR: virtual environment for archaeological artefacts restoration

This demo presents a virtual environment for assembling archaeological artefacts from 3D scanned fragments. We have implemented a set of interaction techniques tailored to this specific task, allowing users to examine, manipulate and assemble fragments to obtain the original shape of the object. The tool is developed and continuously tested by domain experts from the field of anthropology. The presented pilot user study confirms our initial expectation that the restoration process using a virtual environments can be significantly faster than restoration done in a desktop environment keeping the same level of assembly precision.