Petra Urbanová | North Carolina State University (original) (raw)

Papers by Petra Urbanová

While assessing skeletal injuries in human skeletal remains, forensic anthropologists are frequen... more While assessing skeletal injuries in human skeletal remains, forensic anthropologists are frequently presented with fractured, fragmented, or otherwise modified skeletal remains. The examination of evidence and the mechanisms of skeletal injuries often require that separate osseous elements be permanently or temporarily reassembled or reconstructed. If not dealt with properly, such reconstructions may impede accurate interpretation of the evidence. Nowadays, routine forensic examinations increasingly incorporate digital imaging technologies. As a result, a variety of PC-assisted imaging techniques, collectively referred to as the virtual approach, have been made available to treat fragmentary skeletal remains. The present study employs a 3D virtual approach to assess mechanisms of skeletal injuries, and provides an expert opinion of causative tools in three forensic cases involving human skeletal remains where integrity was compromised by multiple peri-or postmortem alterations resulting in fragmentation and/or incompleteness. Three fragmentary skulls and an incomplete set of foot bones with evidence of perimortem fractures (gunshot wounds) and sharp force trauma (saw marks) were digitized using a desktop laser scanner. The digitized skeletal elements were reassembled in the virtual workspace using functionalities incorporated in AMIRA ® version 5.0 software, and simultaneously in real physical space by traditional reconstructive approaches. For this study, the original skeletal fragments were substituted by replicas built by 3D printing. Inter-method differences were quantified by mesh-based comparison after the physically reassembled elements had been re-digitized. Observed differences were further reinforced by visualizing local variations using colormaps and other advanced 3D visualization techniques. In addition, intra-operator and inter-operator error was computed. The results demonstrate that the importance of incorporating the virtual approach into the assessment of skeletal injuries increases with the complexity and state of preservation of a forensic case. While in relatively simple cases the virtual approach is a welcome extension to a traditional approach, which merely facilitates the analysis, in more complex and extensively fragmentary cases such as multiple gunshot wounds or dismemberment, the virtual approach can be a crucial step in applying the principles of gunshot wounds or sharp force traumatic mechanisms. The unrestricted manipulation with digital elements enabling limitless repairs and adjustments to a " best-case scenario " also produced smaller inter-operator variation in comparison to the traditional approach.

The present paper aims to test performances of semi-automatic tools for mesh-to-mesh processing w... more 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.

Creating digital replicas of unique biological findings or archeological artifacts has become a d... more Creating digital replicas of unique biological findings or archeological artifacts has become a desirable task, which enables to spare original integrity and enhance accessibility of valuable objects to a wide range of experts as well as public. In recent years, specialized scanning devices have been challenged by performance of photogrammetry software tools capable of processing unstructured image sets and providing three-dimensional digital models in return. Simplicity, portability and affordability predetermine photogrammetry to be the method of choice if three-dimensional documentation is to be conducted at remote facilities and outdoor locations. The present paper tests technical limitations of two 3D documentation techniques – close range photogrammetry carried out in Agisoft PhotoScan software and laser scanning conducted with MicroScribe/MicroScan scanning unit while documenting pelvic bones and sacra from the Upper Paleolithic triple burial of Dolní Věstonice, Czech Republic. For photogrammetry, two different approaches to generate closed textured 3D models were confronted – alignment of partial polygonal meshes and joint processing of multiple image sets. Our results showed that photogrammetry provided high-resolution 3D models appended by photorealistic texture. In terms of depicted details, the photogrammetry-generated models were comparable to those of laser scanning. However, the robust performance of the employed algorithm was achieved at the expense of extensive time and labor demands, which for many experts may be difficult to justify. In conclusion, photogrammetry should be considered a suitable substitute for surface scanners only if conducted for occasional and/or out-of-lab documentation tasks. KEY WORDS: Photogrammetry ‒ Laser scanning ‒ Three-dimensional documentation ‒ Skeletal remains ‒ Paleoanthropology ‒ Dolní Věstonice ANTHROPOLOGIE

Generating large human crowds of distinguishable individuals is one of the challenges in the gami... more Generating large human crowds of distinguishable individuals is one of the challenges in the gaming industry. When the scene contains many characters, it becomes impracticable to create all the individual characters manually. However, the requirement for the different appearances of individuals in a crowd, namely their faces, is now in greater demand. Therefore, this paper describes our solution to the automatic generation of human faces that are created as a composite of facial parts of 3D scans of real human faces. However, the user has the possibility to further adjust the composite by designing replacements, leading to a desired appearance. The final composite can be exported and attached to a given avatar. To evaluate the usability of our solution, we performed two case studies. The conducted perception study performed with 104 participants aimed to confirm the decreasing human ability to recognize morphologically modified faces. The morphological study focused on the quantification of the extent of facial modifications. Both studies were performed by domain experts from psychology and anthropology.

Background and aim: The effect of socioeconomic factors (living conditions) and parental smoking ... more Background and aim: The effect of socioeconomic factors (living conditions) and parental smoking habits on development of facial morphology and body measurements was studied on a longitudinal Czech sample of 25 girls and 25 boys.
Subjects and methods: A set of studied digitalized photographs taken from 0.5 to 18 years in intervals of 6 months originated in the Brno Longitudinal Study. Facial shape changes of sub-adult participants were described using a configuration of 27 landmarks and further studied by using methods of geometric mor-phometric and multivariate statistics. In order to localize growth-related shape changes within the face, the studied region was divided into upper, middle and lower facial units and analyzed separately.
Results and conclusion: The results show that in the course of ontogenesis there is a strong correlation between facial shape change and body measurements, height included (r=0.10 and r=0.24 in boys and in girls, respectively). The pubertal spurt of the facial shape change rate was revealed at 10.5 years in girls and at 11.5 years in boys. The earlier onset of the pubertal rate increase in facial shape changes in boys was associated with records of poor living conditions. In addition, the mother's smoking habits were linked to a noticeable facial shape change.

Facial image identification is an area of forensic sciences, where an expert provides an opinion ... more Facial image identification is an area of forensic sciences, where an expert provides an opinion on whether or not two or more images depict the same individual. The primary concern for facial image identification is that it must be based on sound scientific principles. The recent extensive development in 3D recording technology, which is presumed to enhance performances of identification tasks, has made essential to question conditions, under which 3D images can yield accurate and reliable results. The present paper explores the effect of mesh resolution, adequacy of selected measures of dissimilarity and number of variables employed to encode identity-specific facial features on a dataset of 528 3D face models sampled from the Fidentis 3D Face Database (N $ 2100). In order to match 3D images two quantitative approaches were tested, the first based on closest point-to-point distances computed from registered surface models and the second grounded on Procrustes distances derived from discrete 3D facial points collected manually on tex-tured 3D facial models. The results expressed in terms of rank-1 identification rates, ROC curves and likelihood ratios show that under optimized conditions the tested algorithms have the capacity to provide very accurate and reliable results. The performance of the tested algorithms is, however, highly dependent on mesh resolution and the number of variables employed in the task. The results also show that in addition to numerical measures of dissimilarity, various 3D visualization tools can be of assistance in the decision-making.

The accurate determination of the sex of human skeletal remains is a critical challenge in forens... more The accurate determination of the sex of human skeletal remains is a critical challenge in forensic pathology and skeletal anthropology. The pelvis and the skull are the most commonly used skeletal sites for determining the sex of skeletons. In the skull, the supraorbital region, which includes the supraorbital margin, is considered a specific sexually dimorphic trait. In the traditional approach, sex is determined through visual and tactile assessment. This paper introduces a methodology for the objective quantification of sexually dimorphic features using wavelet transform, which is a multiscale mathematical tool that allows for the measurement of shape variations that are hidden at different scales of resolution. The method was successfully applied for the sex determination of a pilot sample of 3D meshes—digital records of supraorbital morphology. This information can be used by experts to improve the accuracy of biologic profile assessment of a human skeleton, and to describe the geographic and temporal variations within and among populations.

The accurate determination of the sex of human skeletal remains is a critical challenge in forens... more The accurate determination of the sex of human skeletal remains is a critical challenge in forensic pathology and skeletal anthropology. The pelvis and skull are the most commonly used skeletal sites for determining the sex of skeletons. In the skull, the supraorbital region, which includes the supraorbital margin, is considered a specific sexually dimorphic trait. In the traditional approach, sex is determined through visual and tactile assessment. The present paper introduces a methodology for the objective quantification of sexually dimorphic features using Wavelet Transform, which is a multi-scale mathematical tool that allows for measurement of shape variations that are hidden at different scales of resolution. The method was successfully applied for sex determination of a pilot sample of 3D meshes – digital records of supraorbital morphology. This information can be used by experts to improve the accuracy of biologic profile assessment of a human skeleton, and to describe the geographic and temporal variations within and among populations.

Forensic pathologists ought to take photographs of all important external and internal findings r... more Forensic pathologists ought to take photographs of all important external and internal findings revealed at
autopsy. Tethered photography offers a sophisticated strategy for direct visualization of captured images in an autopsy room, either wired or wirelessly, on larger screens (e.g., smartphones, tablets, monitors). It involves transmitting captured images from a digital camera to a computer or a mobile device, where the image files are saved to a hard drive and displayed on the device’s screen. We present a simple technique for tethered photography suitable for an autopsy room environment based on using memory cards that can create an ad hoc Wi-Fi network to which a preferred screen device can be then connected. Wireless transfer of image files according to their size showed delays of seconds. The maximum distance between a camera and a selected device for effective transmission of the image files was established as 20 meters.

Three-dimensional surface technologies particularly close range photogrammetry and optical surfac... more 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 PhotoScan1 software and stereophotogrammetry-
based Vectra H11, 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 Kra´ love´ , 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 3608 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 20 min, the post-processing was much more time-demanding and required up to
10 h 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.

It is important that forensic pathologists are familiar with variations in the size and shape of ... more It is important that forensic pathologists are familiar with variations in the size and shape of the laryngohyoid complex when interpreting injuries and pathology of the head and neck region. Accurate postmortem examination of the laryngohyoid structures may be difficult if anatomical variation in these structures is present. Agenesis of the upper horns of the thyroid cartilage has medicolegal significance because it may be mistaken for a fracture or other trauma-related conditions. We present 3 cases with different forms of agenesis of the superior cornu of the thyroid cartilage, namely, right unilateral, left unilateral, and bilateral agenesis.

Traditionally, the human hand has been analyzed by measuring and comparing individual segments (f... more Traditionally, the human hand has been analyzed by measuring and comparing individual segments (fingers, phalanges, metacarpals) without considering their mutual spatial relationships. The present study aimed to analyze intra-population variability of the human hand form in the region of fingers as whole with special focus given to sexual dimorphism and the association between shape and size. Right hands of 99 females and 70 males, mostly college students, were scanned with a document scanner in standardized position from the palmar side. For each image 2-D coordinates of 16 landmarks were recorded and variability between configurations of landmarks were studied using geometric morphometrics. To understand patterns of variability, shape spaces were decomposed into affine and non-affine subspaces and further studied separately. The prevalence of the total variability was associated with affine shape change identical for the whole studied region of the hand. Its major portion was represented by shearing in proximodistal direction and to a lesser extent by straining in the relative hand width. The strain also strongly correlated with size of the hand. Intersexual differences were represented by affine change in which the proximodistal shearing was inextricably tied with the differences in relative hand width. Local non-affine sex differences were found in specific phalanges and might be associated with differences in finger length ratios. We presume that separating global and local sexually dimorphic features of the hand might shed light on the origin/onset of the dimorphism during early ontogeny – the more local the feature, the later differentiated, and thus influenced by different ontogenetic factors than global features.

Conference: Mikulov Anthropology Meeting 2014, At Mikulov, Volume: The Dolní Věstonice Studies, vol. 20

Three-dimensional recording of the human face or other parts of the human body has recently gaine... more Three-dimensional recording of the human face or other parts of the human body has recently gained increasing importance in a variety of clinical, biomedical and biological research fields, anthropology included. The vast selection of 3D surface imaging devices available nowadays was preceded by the development of 3D imaging technology in the 1970s and 1980s; at that time it was mostly focused on industrial applications, such as quality control aimed at evaluating potential defects and inconsistencies between a design and its final product . Since the early 2000s, imaging devices designed specifically for capturing the human face and body have emerged . Recording the human body is specific and differs from collecting data from other static objects. In living humans, total immobilization is hardly achievable. Therefore, in order to avoid any distortions caused by movements the recording time has to be very short. divide devices for capturing human body surface into several classes according to the technology they are based on: stereo-photogrammetry, lasers, structured light, video-imaging, radio sources and others. The final digital output is typically a 3D mesh, otherwise known as a polygonal model, representing the morphology of the scanned object with or without surface color or texture. The texture gives the virtual model realistic appearance by assigning color and other optical qualities to each polygon of the polygonal model. Traditionally, quantitative anthropological examination of the human face is based on direct measurements. Since its discovery and throughout the entire 1900s analog and later digital photography were widely used for documentation and analytical purposes, i.e., indirect anthropometry (Farkas 1994; Preedy 2012). The development of 3D scanning technologies expands the traditional framework by allowing the analysis of quantitative data to be carried out on a realistic three-dimensional virtual replica capturing the original geometry as a whole without substantial data reduction and compromises (Weber and Bookstein 2011). As one of many benefits, threedimensional digital outputs are exploitable by both traditional and advanced analytical techniques, such as geometric morphometrics, and allow quantification of the form (shape) of objects in a complex manner (Bookstein 1991; Rohlf and Bookstein 1990; Slice 2005). These 3D models are also easily editable and modifiable by various software tools or transformable back to the physical reality by means of 3D printing. In addition to being measured, the human facial morphological variation is conventionally approached through visual assessment and description of non-metric traits, also known as anthroposcopy Farkas 1994). Within the framework of forensic sciences unique facial traits and features are employed in the process of personal identification. In order to facilitate the process of face description several standardized descriptive and classification systems have been developed . Usually, they take the form of a guidebook or an atlas mapping out facial markers and featuring drawings or photographs complemented by short verbal descriptions. Recently, such guidebooks have been impacted by the expansion of 3D imaging devices by implementing 3D virtual models as interactive charts. There are many areas of basic and applied research involving 3D face processing that could benefit from access to a 3D face model database. Several global databases have been made available; for instance, the Texas 3D Face Recognition Database (Gupta et al. 2010) and the Bosphorus Database (Savran et al. 2009), both primarily serving the purpose of face recognition or facial expression analysis. Building a reference database, albeit time-and energy-consuming, confers crucial advantages: full control over data collection and unrestricted access to them. The present paper introduces the 3D Virtual Model Database of Human Faces and its practical application at The Dolní Věstonice Studies, vol. 20 -Mikulov Anthropology Meeting 2014 178 3d VIrtual Model database of HuMan faces Figure 1. A schematic illustration of the steps involved in creating the entire face model captured by Vectra M1. From top to bottom: Vectra M1; raw partial models; edited models prepared for merging; merged model without texture (left) and with texture (right).

Assessing of archaeological features, such as settlement and grave pits, is traditionally a matte... more Assessing of archaeological features, such as settlement and grave pits, is traditionally a matter of careful and sometimes time-consuming examination by using comparative methods. Despite considered one of fundamentals in archaeology this approach is known to be heavily dependant on a degree of researcher’s erudition and his/her subjective perception of studied issues. Therefore, as alternative, statistics-based methods were suggested as helpful assistance while dealing with unspecified features. This paper re-examines 57 unspecified features from the Southern Suburb of the Great Moravian stronghold Břeclav-Pohansko, whose purpose was previously left unresolved. In order to distinguish grave pits from settlement features we combined the traditional comparative assessment approach with computer-aided techniques featuring univariate and multivariate statistics. The comparison between the set of unspecified features and true grave pit originated from the same archaeological site was carried out on features’ dimensions and orientation.

Legal Medicine

In the framework of forensic anthropology osteometric techniques are generally preferred over vis... more 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 percent 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% to 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.

URBANOVÁ, P.; HEJNA, P.; ZÁTOPKOVÁ, L. & SAFR, M. The asymmetry and modularity of the hyoid bone.... more URBANOVÁ, P.; HEJNA, P.; ZÁTOPKOVÁ, L. & SAFR, M. The asymmetry and modularity of the hyoid bone. Int. J. Morphol., 32(1):251-260, 2014.

In order to develop an identification key for distinguishing between human and non-human osteolog... more In order to develop an identification key for distinguishing between human and non-human osteological samples, bone structure of several animal taxa was studied using quantitative microscopy. Both domestic and wild species were included in the sample, analysed at both micrometric (13 variable parameters) and macrometric (3 variable parameters) scales. The observed data was first used to evaluate inter-and intra-species diversity. Least determinative parameters were then eliminated via stepwise discriminant function analysis. The most discriminating micrometric properties of compact bone tissue appeared to be: number of osteons in 1 mm 2 , maximum osteon diameter, maximum diameter and area of Haversian canal, and mid-shaft femoral cortical thickness. Ultimately, two different equations for discriminating between human and non-human bone are formulated. The first type uses only histometric properties of bone structure. The second type of equations combines histometric measurement and grossest morphometric parameter -cortical thickness. The latter equations correctly predict taxonomic classification in 100% of cases.

The accurate determination of the sex of human skeletal remains is a critical challenge in forens... more The accurate determination of the sex of human skeletal remains is a critical challenge in forensic pathology and skeletal anthropology. The pelvis and skull are the most commonly used skeletal sites for determining the sex of skeletons. In the skull, the supraorbital region, which includes the supraorbital margin, is considered a specific sexually dimorphic trait. In the traditional approach, sex is determined through visual and tactile assessment. The present paper introduces a methodology for the objective quantification of sexually dimorphic features using Wavelet Transform, which is a multi-scale mathematical tool that allows for measurement of shape variations that are hidden at different scales of resolution. The method was successfully applied for sex determination of a pilot sample of 3D meshes – digital records of supraorbital morphology. This information can be used by experts to improve the accuracy of biologic profile assessment of a human skeleton, and to describe the geographic and temporal variations within and among populations.

Conference: Mikulov Anthropology Meeting 2014, At Mikulov, Volume: The Dolní Věstonice Studies, vol. 20

Legal Medicine

Recent advancements in three-dimensional face scanning technology and image processing have broug... more Recent advancements in three-dimensional face scanning technology and image processing have brought novel opportunities into many research and applied areas, forensic sciences included. While benefiting from the presence of depth information and relative independence to illumination and viewpoint forensic evidence based on 3D technology remains in many cases affected by uncontrolled and unpredictable behavior and appearance of recorded subjects. Of the known interfering factors (hairstyle, occlusion etc.), face expression variance may particularly possess major obstacles for robust 3D face recognition procedures.

In the field of forensic sciences, morphology of the hyoid bone provides important evidence on vi... more In the field of forensic sciences, morphology of the hyoid
bone provides important evidence on victim’s biological
profile, e.g., sex, ancestry, age at death and on cause of
death where various neck or head related injuries can
be manifested as hyoid fractures. In regards to their
mechanisms fractures of hyoids have been associated with
accidental traumas (e.g., car accidents, falls), self-inflicted
(e.g., hanging) or assaulted injuries (e.g., manual or ligature
strangulation). In addition, postmortem hyoid fractures
often occur when a larynx is being harvested and examined
at autopsy which may easily lead to misdiagnoses.
Previous studies indicated that a variety of factors (age[1],
sex[2], mechanism of injury[3]) might be instrumental in
occurrence of hyoid fracture. Hyoid morphology, in particular,
has been suggested as helpful while assessing mechanism
of skeletal injuries.

Three-dimensional data has become increasingly employed in assessment of human remains of unknown... more Three-dimensional data has become increasingly employed in assessment of human remains of unknown origin. In the framework of forensic anthropology the most advanced anthropological tools available nowadays process discreet osteometric points, more specifically their 3D Cartesian coordinates. Although this approach respects high requirements on reliability and repeatability it omits substantial amount of biological information encoded in the studied morphology. In contrast, recently developed tools for direct 3D mesh-to-mesh comparison have potential to bridge the gap between traditional trivial and computer-aided techniques allowing quantifying skeletal morphology in a very complex manner.
Potentials of mesh-based tools for sex and ancestry assessment were explored on a basis of a sample of 40 human crania which originated in two documented Brazilian collections (USP and UNIFESP, São Paulo, Brazil). The cranial morphology were 3D digitized via photogrammetry and laser scanning and further quantified in freely available software FIDENTIS Analyst (www.fidentis.com) combined with k-means cluster analysis. The program quantifies dissimilarity in the sample pairwise on a basis of closest vertex-to-vertex distances. This global measure of differences is further reinforced by visualizing local variations using color scalar maps. The acquired results were confronted with those provided by 3D-ID software (www.3d-id.org), which enables performing sex and ancestry assessment by processing Cartesian coordinates of selected craniometric points.
The present approach allowed us to describe within-sample morphological variation, manifestation of sexual dimorphism and to specify population-related traits. At the present state of development, it remains unclear to what extend 3D model attributes may affect final diagnoses or whether the adopted algorithm is applicable on other skeletal regions. The surface- or mesh-based techniques , however, proved to be potentially very perspective for assessing those cranial regions which generally lack natural, biologically homologous landmarks as well as for developing automated or semi-automated diagnostic systems.

For the last few years several projects have invested into establishing image-guided postmortem e... more For the last few years several projects have invested into establishing image-guided postmortem examination in the framework of forensic pathology – utilizing either medical imaging technologies (CT, MRI) or optical surface scanning. Cross-sectional and surface data are both stand-alone techniques which bring a number of benefits but also few shortcomings into postmortem examinations. While CT-based images have the capacity to generate 3D model of any internal structure, under the condition that the contrast between two neighboring tissues or structures is sufficient, they provide little or no information about surface coloring (bruises, lesions), spots, or subtle morphological interferences. Optical surface models, in contrast, lack information going underneath the surface, but feature high resolution geometry and surface color information.
Three-dimensional surface technologies, in particular photogrammetry, optical and laser scanning have recently advanced into affordable, flexible and accurate techniques involving a reasonably long learning curve. Forensic postmortem investigation as performed on a daily basis, however, has not benefited from their full potentials and the currently utilized methods are far behind of what 3D surface documentation can provide. Conventionally, two-dimensional photography presents the gold standard utilized throughout the entire process of postmortem examination, documenting corpse’s pre-autopsy state of preservation, presence of unique somatic traits or perishable findings, injuries and/or pathological changes. Photography, however, discards surface depth – information highly valuable in terms of damage assessment or trait uniqueness.
In the present paper we tested two approaches to 3D external body documentation – traditional digital camera-based photogrammetry combined with commercial Agisoft PhotoScan® software and stereophotogrammetry-based scanner VECTRA H1®, a novelty product among portable handheld surface scanners. In order to conduct the study two forensic cases admitted for postmortem examination at the Department of Forensic Medicine, Hradec Králové, Czech Republic were selected. A 63-year-old male, death due to traumatic, self-inflicted, injuries (suicide by hanging), and a 63-year-old male, diagnosed with heart failure. Both cases were photographed in 360° manner with a digital camera mounted on a tripod. Altogether 35 to 70 images per case were taken, corresponding to 20 to 40 minutes of capturing time. Approximately the same time was required to document the body surface with the handheld scanner, where up to 120 scans were taken.
The optical surface scanner proved itself to be a useful tool for enabling to document small-to-large areas of the body surface. As not being specifically designed to scan objects on a larger scale post-processing requires rather time-consuming manual image alignment. The device, however, produced high-resolution 3D images, comparable in quality to any professional digital camera. The photogrammetry also provided photorealistic records of body surface capable of capturing, for instance, ligature marks, tattoos, and skin lesions in high quality. Moreover, the utilized software was able to create the whole body 3D surface mesh automatically. Both methods failed, however, when the surface was covered with body hair and reflective, moist areas were being documented.
In conclusion, both methods produce realistic, real sized or easy-to-calibrate 3D surface models useable as an advanced method of postmortem documentation and easily fit to be the subject of further examinations, such as 3D mesh comparison or indirect measurement (e.g., body measurements, angles of penetrations). Digital 3D data can be easily archived, transported and shared between laboratories; they provide a real-time access for re-examination of physical evidence. Furthermore, no requirements for calibration or operating skills make them ideal to be easily integrated into a daily workflow.

The increasing accuracy of sensors used for capturing human faces together with advances in compu... more The increasing accuracy of sensors used for capturing human faces together with advances in computer graphics and vision have had a great impact on the quality of images that are being admitted for forensic examinations. Matching image-based forensic evidence from crime scenes (surveillance videos) with suspects (mug shots, ID photographs) is known to be highly dependent on the image quality, optical distortion, perpetrator’s head position or disguise. Therefore, in many forensic cases automatic computer-aided systems for facial recognition are of little use as they fail to provide positive (i.e., to identify a person), but also negative (i.e., to eliminate a person) results. Currently, new development in 3D technologies has offered solutions to overcome these key limits by allowing 2D to 3D and 3D to 3D digital record matching.
Despite that fact that 3D video surveillance systems are being developed and are expected to be widespread in the near future so far there has not been sufficient scientific evidence on how accurate and reliable it is to utilize three-dimensional records of human face for the purpose of forensic identification or whether matching 3D digital evidence meets legal requirements that demand the use of quantifiable scientific methods with known error rates. 3D human faces are generally captured by a variety of in-lab equipment – laser or stereophotogrammetry-based devices, which eventually generate plain or textured photorealistic 3D meshes. Processing and matching two or multiple 3D surface meshes is a challenging task which requires a proper alignment algorithm together with numerical and visual representations. There is a number of existing algorithms and metrics, such the Iterative Closest Point (ICP) for aligning two surface datasets or Haussdorff distances for quantifying actual distances between two aligned meshes. In addition, surface-based visualization technique is commonly performed using colormaps and scalar fields.
In the present paper we tested the currently available algorithms for matching 3D faces using a dataset sampled from a large database of high resolution 3D scans - the 3D Virtual Model Database of Human Faces. Established at authors’ home institution and in a stripped version available throughout the project web site the database contains approximately 2,000 scans acquired for a variety of reasons - studies of individual and age-related human variation, growth and development studies, parent-to-offspring resemblance etc. Specifically, it contains both sub-adult and adult individuals scanned repeatedly under controlled and semi-controlled conditions as well as those composing a large reference sample of non-matching individuals at a population level.
To improve the matching process we modified available algorithms by introducing an iterative process of successive improvement of the average face sped up with KD-trees which allows a better alignment of 3D meshes when a multiple comparison is launched. The magnitude of dis/similarity among aligned meshes was expressed by signed and absolute Haussdorff distances, i.e., the greatest of all the distances from a vertex in reference mesh to the vertex in the compared one and further extended to incorporate vertex to polygon and inter-polygon distances.

The displayed facial reconstructions of uncovered skeletal remains are courtesy of Dr. Paulo Miam... more The displayed facial reconstructions of uncovered skeletal remains are courtesy of Dr. Paulo Miamoto and Cícero Moraes. The authors are grateful for their assistance and collaboration.

OUTLINE the purpose of this presentation is to to present a method which is developed entirely on... more OUTLINE the purpose of this presentation is to to present a method which is developed entirely on principles of shape analysis (geometric morphometrics) in conjunction with 3D reference database these principles advance estimating individual's biological sex and ancestry in skeletal remains to demonstrate its suitability for the U.S. and non-U.S. forensic settings WHEN TRADITIONAL MORPHOMETRICS FAILS… Two crania of different morphology 6 inter-landmark distances corresponding to standard cranial measurements

The studied sample (N =367) consisted of three sub-samples which originate from renown identifi e... more The studied sample (N =367) consisted of three sub-samples which originate from renown identifi ed osteological collections, two of them of the Portuguese origin and a single coming from the Czech Republic. For each specimen, individual's sex, age and cause of death was documented.

Ladislava Horáčková and Petr Dubový for providing the privilege of studying the