Forensic Science International 303 (2019) 109943

Contents lists available at ScienceDirect

Forensic Science International

journal homepage: www.elsevier.com/locate/forsciint

Technical Note

Virtual reality and integrated crime scene scanning for immersive and
heterogeneous crime scene reconstruction
Jinming Wanga,b,1, Zhengdong Lia,b,1, Wenhu Hua,c , Yu Shaob , Liyang Wangd ,
Rongqi Wud , Kaijun Mad, Donghua Zoua,* ,1, Yijiu Chena,* ,1
a
Shanghai Key Laboratory of Forensic Medicine, Academy of Forensic Science, Ministry of Justice, 1347# West Guangfu Road, Shanghai, China
b
Department of Forensic Medicine, School of Basic Medical Sciences, Fudan University, 130# Dongan Road, Shanghai, China
c
School of Forensic Medicine, Southern Medical University, 1838# Guangzhou Avenue North Road, Guangzhou, China
d
Shanghai Key Laboratory of Crime Scene Evidence, Institute of Forensic Science, Shanghai Public Security Bureau, Shanghai, China

A R T I C L E I N F O A B S T R A C T

Article history: Crime scene reconstruction plays a significant role in crime solving by helping to determine the course of
Received 3 December 2018 events. Non-invasive, high-resolution measurement and increased insight are always the goal of forensic
Received in revised form 27 July 2019 crime scene documentation. However, entire crime scenes cannot be effectively reconstructed with
Accepted 26 August 2019
traditional methods. In this study, we present a portable system that consists of a laser scanner, two
Available online 13 September 2019
hand-held structured light scanners and a low-cost virtual reality (VR) headset with a mobile power
supply to conduct multi-angle and omnidirectional three-dimensional spatial data collection of crime
Keywords:
scenes. To demonstrate practical use, a real case has been analysed to verify the feasibility and
Forensic science
3D scanning
effectiveness of the system.
Virtual reality The system accurately obtains information on decedent injuries, possible injury-inflicting tools and on-
Crime scene reconstruction site traces. Various types of evidence from the crime scene can be jointly studied by three-dimensional
visualization to develop a cohesive story. The data are presented via immersive VR rather than displayed
on computer screens. The relationship between evidence chains enables us to achieve a complete crime
scene reconstruction, using the specialized knowledge of experts and computer-aided forensic tools to
analyse the causes of damage and identify suspects. The use of three- dimensional imaging techniques
allows a more insightful survey and several useful analyses, such as accurate measurement, relative blood
source location determination and injury-inflicting tool comparison.
© 2019 Published by Elsevier B.V.

1. Introduction limiting further reconstruction. In addition, most on-site records in

Investigation of violent incidents is challenging and involves
various kinds of specialized knowledge in different subjects. Every
crime scene investigator has an individual perspective, and even
the most diligent will make mistakes at times, especially in a
complex scenario. If potential evidence is not recognized
immediately, it could easily be altered. In China, some public
security bureaus at the local level lack the ability to manage
complex cases; therefore, they seek help from experts in the field
to perform a second survey. Nevertheless, in some cases, the initial
investigation was careless, and some details were altered, thus
* Corresponding authors at: Shanghai Key Laboratory of Forensic Medicine,
Academy of Forensic Science, Ministry of Justice, 1347# West Guangfu Road,
Shanghai, 200063, China.
E-mail addresses: zoudh@ssfjd.cn (D. Zou), cyj1347@163.com (Y. Chen).
1
These authors contributed equally to this work.
traditional crime scene investigations rely on photographs and
manual sketches. Such photographs and reported descriptions can
thoroughly overwhelm laypeople, such as the jury, in a court
setting. Consequently, a more objective, precise and comprehen-
sive solution for crime scene documentation is urgently needed.
Due to the advantages of non-invasive and rapid measurement,
high precision and a large quantity of information provided, 3D
imaging techniques such as laser scanning and structured light
scanning have proven efficient for many forensic applications in
recent years, such as document surfaces in forensic pathology and
clinical forensic medicine [1–5], bite mark analysis [6,7], face
recognition [8] and incident scene reconstruction [9–12]. However,
most scanners lack mobility and only trained specialists can use
this method skillfully, resulting in the time-consuming of
scanning. Besides, the final 3D visualization is always presented
on a bidimensional computer screen. This method suffers from a
fatal defect: the method of representation may create bias, in
contrast to a real-life visit. Virtual reality (VR) is a potential

http://dx.doi.org/10.1016/j.forsciint.2019.109943
0379-0738/© 2019 Published by Elsevier B.V.
2 J. Wang et al. / Forensic Science International 303 (2019) 109943
solution to provide a better alternative to crime scene visitation
[13]. Moreover, the body of the deceased, the accident scene
including all traces and possible injury-inflicting objects are
equally important in a virtual 3D reconstruction. It should be
emphasized that none of these 3D techniques—laser scanning,
surface scanning or digital close-range photogrammetry—could be
applied separately in forensic documentation.
In this article, we demonstrate an integrated 3D scanning
platform involving laser scanning and structured light scanning
techniques to obtain all potential evidence from the entire crime
scene. The laser scanning data can be visualized in 3D by a low-cost
VR headset, providing an opportunity for users to move through
the virtual scene. All the 3D modelling data can be permanently
preserved and easily transmitted.
2. Materials and methods
2.1. The case
A real case was analysed to verify the feasibility and
effectiveness of the system with the cooperation of the Shanghai
Key Laboratory of Crime Scene Evidence. The incident can be
described as follows. A 30-year-old woman was found unrespon-
sive in an apartment in a lying position on the bed in the bedroom
(Fig. 1a). The doctor arrived first and may have moved the victim.
Then, scientific police arrived at the site and our team followed
soon after. The police report showed that bloodstains were found
on the walls around the bed, bedside table, closet, quilt, and
mattress. Additionally, some footprints were located at the
entrance to the bedroom and in the kitchen. Three items were
notable at the crime scene: a bloodstained cleaver in the kitchen
(Fig. 1d), a broken bottle in the living room (Fig. 1e) and a glass
vase near the bed (Fig. 1f). The police were highly suspicious that
an acquaintance may have committed the crime and thought it
was a deliberate murder. Three-dimensional methods were
applied in this case to collect physical evidence and to verify
the hypothesis.
Fig. 1. Complete model of the crime scene covering three rooms, with footprint locations. (a) victim’s post-mortem body position as photographed by the police; (b) quick
view option in SCENE, which provides a more vivid view and has measurement capabilities; (c) detailed scan of a footprint by the structured light scanner; (d) 3D model of the
cleaver; (e) broken wine bottle in the living room; and (f) 3D model of the glass vase near the bed.
The tasks performed by our team consisted of several parts,
including integrated scanning, medico legal evaluations, blood-
stain pattern analysis, footprint analysis, final data collation and VR
visualization. A flow chart (Fig. 2) details the entire process and
analytical procedure.
2.2. 3D scanning platform: documentation of all evidences from crime
scene
2.2.1. The overall scene
The overall scene was created using Faro Focus3D S120 (Faro,
USA) with a range accuracy of 2 mm. The scanner is a fully portable
device that provides a distance range from 0.6 to 130 m. The laser
scanner measured all surfaces over 360 in the horizontal direction
and 305 in the vertical direction, generating a circular hole below
the tripod in a single scan. Therefore, we performed several scans
from different locations were performed to improve the accuracy
and eliminate the holes in the scans. Then, those point clouds were
stitched together by using the known locations of the ground
control points (GCPs) or other features within the scene. To
complete scanning more quickly and allow for a real-time display
of the scene, we used the field instant method; then, we can
continue with additional scans as needed to supplement incom-
plete coverage. Three overall laser scans captured the entire crime
scene in the bedroom, whereas 2 scans captured the living room,
and 2 scans captured the kitchen.
2.2.2. External body
External documentation of the corpse was conducted in the
autopsy room. We removed the clothes and cleaned the body of
decreased to exposure the injuries clearly, and then we performed
the scan. Given its larger field of view, the forensic pathologist
performed whole-body scanning by Go! Scan50 (Creaform,
Canada) with a resolution of 1 mm before the autopsy. The
injuries, such as scalp lacerations, a stab wound and parallel
incised wounds, were intensively scanned by Go! Scan20 (Crea-
form, Canada) with a resolution of 0.1 mm. We performed one front
 J. Wang et al. / Forensic Science International 303 (2019) 109943
Fig. 2. Flow chart showing the equipment and procedures used for the technique.
body scan, one back body scan and four detailed scans for injuries.
Then the whole-body model and injury models were converted
into an overall model. These hand-held structured light scanners
enable users to observe the state of the scanned model in real time
by connecting to a laptop and to augment any lacking areas with
additional scanning. A detailed and textured 3D model was created
quickly as the scanner moved over the deceased.
2.2.3. Crime scene objects and other evidence
The involved objects were digitised in 3D except for the broken
wine bottle. We scanned the bloodstained cleaver and the glass
vase using Go! Scan50 with a resolution of 1 mm. The bloodstains
on the surface of the glass vase were recorded by Go! Scan 20 with
a resolution of 0.1 mm for further analysis. We also recorded 35
footprints via 11 surface scans using Go! Scan50 with a resolution
of 1 mm. The distribution of the footprints was recorded by overall
scanning. The laser scanning resolution was insufficient for
bloodstain pattern analysis. Thus, twelve high-resolution photo-
graphs were captured, and the shoot angle required to be
perpendicular to a plane, were used for further blood pattern
analysis.
2.2.4. VR
We used the HTC VIVE Pro (HTC, Taiwan) VR headset in this
system. This equipment consists of a headset, a pair of hand-held
controllers and a positioning-system tracking display. The
headset covers a nominal field of view of approximately 110
with a resolution of 1440  1600 pixels per eye and 2880  1600
pixels for the two-eye display. Its tracking technology uses two
3
laser emitters called Lighthouses. These two boxes emit
alternating horizontal and vertical infrared laser sweeps spanning
120 in each direction [14], which allow the user to move within a
certain scope and interact with the virtual scene using the hand-
held controllers. The VIVE Pro allows physical movement within
an area of 6 m  6 m, and operators can move freely by using a
wireless connection.
2.3. Post processing of the data and fusion
All data processing was performed on a laptop with an Intel(R)
Xeon CPU clocked at 2.90 GHz and 64 GB of RAM. Data processing
included initial processing to clean redundant data and data
integration to produce a complete model.
For the laser scanning data, noise, outliers and duplicated
points were automatically detected and removed in SCENE 7.1
(Faro, USA). Then several laser scans from different locations were
converted to an overall scene model. The process of registration
was automatic according to intrinsic character features in scans.
VXelements 6.1(Creaform, Canada) was used for calibration and
configuration of the hand-held structured light scanners and for
collecting data. If the scanner was colliding in transportation or
had not been used for a long time, a calibration was required before
scanning. The software automatically cleared the grid of over-
lapping points, filled in missing data by adding objects and enabled
exportation of the file to other programs, including Geomagic (3D
systems, USA), among others. Finally, we used the open-source,
extensible mesh processing system MeshLab to further process the
polygons of all models for visualization.
4 J. Wang et al. / Forensic Science International 303 (2019) 109943
2.4. The analysis
2.4.1. Bloodstain pattern analysis (BPA)
High-resolution photographs were imported and scaled to the
3D point cloud in Faro Zone 3D software (Faro, USA). A point cloud
should have registration targets for alignment that match the
targets in the photo. The operator marked three pairs of
corresponding points between them for preliminary alignment,
and then an automatic procedure was conducted for high-
precision alignment, which is important because accurate align-
ments are necessary for accurate conversion points.
The length of major and minor axes of the elliptical bloodstain
indicates the angle of impact and directionality of the bloodstain.
The software uses image recognition techniques to approximate
the ellipse when the operator clicks on the bloodstain. These works
were conducted by a BPA expert. If adjustments are needed, then
the operator could use ellipse control points to fit the outline of
bloodstains. The BPA expert could also unselect some bloodstains,
which disrupts the results and flips the direction when the
trajectory line is in the wrong direction. The values of the impact
angle α and the directionality angle g were predicted once the
ellipses were determined. All well-defined bloodstains on walls
and on furniture were accurately measured and then the flight
paths of the bloodstains were visualized in the 3D model of the
crime scene. In this manner, the convergence of straight lines
determines the approximate location of the blood source in the
simulated scene.
2.4.2. Footprints
Each footprint was evaluated by a forensic podiatrist in
VXmodel software (Creaform, Canada). A flowchart of the footprint
analysis is shown in Fig. 3. The central axis was drawn from the
rearmost aspect of the heel of the footprint to the central point of
the second toe. The front and back tangents of the footprint, which
were perpendicular to the central axis, were identified to create the
“footprint length”. The inner and outer tangents, which were
parallel to the central axis, were identified to create the “footprint
breath”. We measured the foot length and foot breadth of 14
footprints, excluding those that were overlapped and broken in
this case. Furthermore, the overlay method [15] was used in
Geomagic software to compare the shape of every footprint to
identify whether the footprint belonged to the same person.
2.4.3. Matching injuries with injury-inflicting tools
The injuries were carefully evaluated by forensic pathologists
during the autopsy. The 3D model of injury-inflicting tools and the
body were imported to MeshLab software. We tried to match the
injuries and damage instruments in the three-dimensional
visualization to test the hypothesis proposed based on medico
Fig. 3. Flow chart of the footprint analysis procedure.
legal evaluation. Bloodstains were also located on the face of the
glass vase, and the presence of blood provided information on the
direction of the spatter. We also analysed these bloodstains using
the methods described in Section 2.4.1.
2.4.4. VR
The overall 3D model was converted into a VR scene in SCENE
7.1 (Faro, USA), which allows users to step into a point cloud
simulated scene with a VR headset. VR presentation was used for a
secondary survey and to show crime scene evidence. In the virtual
crime scene, operators can experience scanning, taking measure-
ments, capturing screens and reading annotations as investigators
do in the field. According to personal preferences and needs, users
can choose different modes of movement (flying or walking) or
transfer to a location of interest from a distance and interact with
the virtual scene using hand-held controllers. One controller
controls the direction of movement in the virtual crime scene, and
the other allows the user to zoom in and measure points of interest.
Screenshots of the virtual world can also be recorded.
3. Results
The complete model of the crime scene, covering three rooms,
is shown in Fig. 1. In this case, 7 overall laser scans and 23
structured light scans were performed for data collection. To
minimise time requirements, we performed surface scanning in
one room while laser scanning was performed in another room.
The entire process of scanning and data processing took
approximately 2 h. The precise time cost of data statistic are
presented in Table 1.
3.1. Medico-legal evaluations
An external surface model of the corpse displayed the locations
of the injuries (Fig. 4). An external examination of the body showed
several scalp lacerations, a stab wound near the superciliary arch
(Fig. 5a), bilateral blepharon-conjunctival haemorrhage and
several parallel incised wounds with an abrasion on the anterior
neck. An internal examination revealed bilateral temporal muscle
haemorrhage. Multiple linear fracture lines were identified in the
right temporo-parietal bone, which extended to the skull base
(Fig. 5b). The brain tissue in the area of the right temporal was
bruised. Further findings included cervical muscle haemorrhage
(Fig. 5c). The forensic pathologist concluded that death was due to
blunt craniocerebral and asphyxia.
The scalp lacerations were irregular in shape and in margin, and
tissue bridges were observed in the lacerations, and contusions
were observed around them. A blunt instrument caused these
injures. The margin of the stab wound near the superciliary arch
 J. Wang et al. / Forensic Science International 303 (2019) 109943 5

Table 1
Time cost of obtaining scan data and data processing.

Type of scan Number of scans Time of single scan Time of initial processing Time of integration
Overall scene 7 8 min 5 min 3 min
The whole body 2 7 min 3 min 2.5 min
Injuries 4 40 s 1 min 1 min
Footprints 11 1 min 1.5 min 1 min
Objects 6 1.5 min 2 min 1 min

Fig. 4. External surface model of the corpse, several scalp lacerations and parallel incised wounds on the anterior neck.

Fig. 5. (a) Stab wound near the superciliary arch, (b) fracture of the right temporal bone, (c) and cervical muscle haemorrhage.

was regular and tissue bridges were not found, thus indicating that
it was a sharp wound. The incised wounds on the anterior neck
only damaged the epidermis. Cervical muscle haemorrhage and
the abrasion in this area highly suggested that the assaulter
attempted manual strangulation.
There were three possible injury-inflicting tools as mentioned
above. Based on the morphology of the injuries, we reasonably
concluded that the stab wound was caused by a broken bottle, the
scalp lacerations were caused by a glass vase, and the incised
wounds were caused by a cleaver. The relationship between the 3D

Fig. 6. Potential 3D matching positions of the injury-inflicting instruments and the head injuries.
6 J. Wang et al. / Forensic Science International 303 (2019) 109943

Table 2
A selection of the results of the bloodstain pattern analysis.

Bloodstain Bloodstain coordinate Ellipse Angle

X(m) Y(m) Z(m) Length(m) Width(m) g(Directionality angle) α(Angle of the inclined plane)
1 3.880 0.274 0.564 0.0061 0.0013 346.4 11.4
2 2.548 0.427 0.755 0.0036 0.0018 85.2 28.4
3 1.573 5.478 1.059 0.0051 0.0038 64.3 48.6
4 1.551 5.342 0.889 0.0023 0.0018 24.8 50.5
5 1.570 5.422 1.050 0.0025 0.0015 61.8 37.2
6 5.467 3.034 0.609 0.0112 0.0025 339.9 13.6
7 5.454 3.304 0.585 0.0061 0.0013 283.6 12.8
8 5.460 3.209 0.825 0.0043 0.0020 296.8 21.3
9 4.475 3.403 0.699 0.0030 0.0025 78.8 58.7
10 2.599 0.433 0.998 0.0074 0.0046 77.6 38.2
11 4.044 3.475 0.523 0.0036 0.0020 273.5 33.2
12 2.840 0.366 0.514 0.0038 0.0025 290.8 39.0
13 4.927 0.799 0.822 0.0051 0.0030 113.6 37.1
14 5.531 2.717 0.705 0.0051 0.0041 214.3 52.7

model of the deceased and that of the injury-inflicting objects was
reconstructed virtually as shown in Fig. 6.
3.2. Bloodstain analysis
Bloodstains were located on the walls around the bed, bedside
table, closet, quilt, and mattress. Based on the analysis of the well-
shaped bloodstains, the analysed centres of the bloodstain origin
were located around the bed, which meant that the incident
occurred on the bed. Certain trajectory results for the bloodstain
patterns are shown in Table 2. In addition, a blood spatter report
was produced by the system to show the detailed results of the
blood spatter trajectories (Fig. 7).
An analysis of the centre of the bloodstains on the glass vase
indicated that the centres were close to one side of the vase, which
suggested that the side face may have been the location of impact.
In addition, the hair present on that side and absence of blood on
the opposite side further confirmed this result (Fig. 8).
3.3. Footprint analysis
Only the footprints of the left foot were found on the spot. The
positions of the footprints in this case are shown in three
viewports, i.e., a 3D viewport of the distribution (Fig. 1), a quick
viewport (Fig. 1b), and a detailed viewport (Fig. 1c), by hand-held
scanners. Measurements could be conducted in all these view-
ports. The average footprint length of 14 complete footprints at the
crime scene was 24.57 cm with a standard deviation of 0.68 cm
(Table 3). According to the unitary regression equation obtained by
the local population (height = 82.889 + 3.739  the average foot-
print length) using this measurement method, we inferred that the
height of the suspect was between 172.25 cm and 177.26 cm. The
length of the footprint indicated that it was more likely to belong to
a male. In the overlay method, the uniform shape suggested that all
footprints belonged to the same person.
The distribution of footprints reflected the movement of their
owner. The opposite directions of the footprints in the doorways of

Fig. 7. BPA results for spatter in the bedroom. The sphere shows the analysed centre of the origin.
 J. Wang et al. / Forensic Science International 303 (2019) 109943 7

Fig. 8. External surface model of the glass vase, with bloodstains on the different faces of the glass vase and hair on one face (d), and analysed centre of bloodstains on the glass
vase (e).

victim ran into the bedroom. Contrary to her wishes, she was
Table 3
Physical measurement of footprints.
caught on the bed and held down by the neck with manual
pressure by the murderer. Then, the murderer picked up the glass
Footprint Footprint length Footprint breath vase, which was nearby, to hit the victim in the head. The victim
1 24.10 8.65 attempted to escape the hit, thus causing the diffuse distribution of
2 25.82 9.44 head injuries. In the second phase, the murderer went to the
3 24.21 8.67
4 24.71 9.41
kitchen for a cleaver to cut the victim's neck to ensure her death. At
5 25.52 9.35 this point, the victim was unconscious and could not move, which
6 25.02 9.32 explains the parallel and dense nature of the incised wounds. In the
7 24.36 9.43 last phase, the murderer left the residence after changing his
8 24.72 8.32
clothes in the living room. All of the above behaviour hinted that
9 25.12 9.93
10 24.17 9.32 the murderer may have been someone familiar with the deceased
11 24.90 8.38 or a person living in the same residence.
12 23.44 8.16 In conclusion, the woman died due to fatal head injuries from
13 23.96 9.12 the glass vase and asphyxia by manual strangulation. The crime
14 23.89 8.23
was likely committed in the spur of the moment, and the suspect
may have been a man living in the apartment.

the bedroom and the kitchen indicated that the murderer made a
return trip from the kitchen to the bedroom. The stained clothes at
the end of the living room were highly suggestive that their owner
was the murderer. Thus, a bold hypothesis emerged that the
suspect had changed his clothes with those drying between the
bamboo poles.

3.4. Reconstruction

The autopsy findings and the BPA results showed that the
glass vase caused the head injuries. A simulation of the strike
action in 3D indicated that the victim may have been in
motion when the vase hit, which was also suggested by the
bilateral distribution of the head injuries. In contrast, the
distribution of the incised wounds on the anterior neck was
parallel and dense. A reasonable explanation was that the
victim may have been unconscious at the time that those
wounds were made. Therefore, we can deduce that the incised
wounds occurred after the head injury. According to the
location of the broken bottle, we reasonably concluded that
the stab wound near the superciliary arch occurred in the living
room. Finally, a VR presentation was used for a secondary
survey and to show crime scene evidence (Fig. 9). Thus, based
on scene and evidence examination, the damage, traces, and
injuries could be linked into a cohesive story of the events of the
case, as follows.
The murderer had a close relationship with the victim so that he
could enter the apartment freely. In the first phase, a squabble
occurred for some reason in the living room, and the murderer hit Fig. 9. Immersive crime scene experience in a display using HTC VIVE Pro, in which
the victim's head with a wine bottle. To avoid further attack, the operators can measure or take screenshots of the virtual world.
8 J. Wang et al. / Forensic Science International 303 (2019) 109943
4. Discussion
Many previous publications have demonstrated how virtual 3D
reconstructions help to determine the course of events of an
accident or crime [11,12,16–18]. Laser scanners, surface scanners
and photogrammetry methods were used to document the victim’s
injuries and items at the crime scene. However, previous studies
have not yet integrated all 3D digital techniques into a platform
that would allow for a joint analysis, and some techniques must
still be improved to meet the needs of real cases.
Time consumption in the field investigations represents an
important factor needing to be considered. The entire process of
scanning and data processing in our system took approximately 2
h. The time required almost equal to that of traditional methods.
Nevertheless, a more complex crime scene and complicated
injuries on the body undoubtedly require more scans and time.
Due to the particularities of crime scenes, the time cost must be
balanced. Fortunately, the quick viewport (panorama) in our
system can compensate for this deficiency to a large extent. On the
other hand, even minimally trained users, such as forensic
pathologists, investigators, forensic podiatrists and BPA experts
in the case, could complete scanning and post-processing alone. All
these factors lead to the platform’s potential applications in crime
scenes.
After the incident occurred, doctors first arrive at the scene and
may have moved the body, and scientific police came subsequently,
followed by our team soon after. After the police extracted the
possible evidence and injury-inflicting tools, we performed the
room scans. Unless such systems are available in local police
stations, it is still difficult to record the data from the original
scene. However, the scanner cost is too high for most institutes.
We recorded all the objects of the crime scene and external
injuries through the overall room and body scan. However, the
resolution of the overall scan may not meet the requirements for
some analyses, such as BPA. To solve the problem, two experienced
investigators and one forensic pathologist performed the detailed
room and body scans respectively. Once the speed, resolution,
quality of the overall scan are improved, additional detailed
scanning will not be required.
In our practical case application, we employed integrated
scanning for the heterogeneous documentation of a crime scene.
Compared to the platform mentioned by Marcin et al. [19], the
scanners in our system combine mobility and scanning speed,
which allow for more measurements to be performed. The laser
scanner can be connected to a computer or tablet via wireless local
area network (WLAN). Users can observe the scan and post-process
it in real time, reducing the need for repeated scans. In addition,
due to the small size, low mass and minimal power use of the Go!
Scan device, this system is suitable for most crime scenes with an
outdoor mobile power source. However, similar to other optical
technologies, the system’s performance is compromised when
light conditions are not satisfactory.
As described in Thomas et al. [20], VR is more reliable for spatial
details than photographs. VR provides a compromise when crime
scenes cannot be physically observed. The advantages of VR
presentation in courts include increased evidence, improved
persuasiveness, accelerated legal proceedings and increased
attention. Ebert et al. [13] described a low-cost immersive VR
system that can simulate the physical presence in a virtual crime
scenario through stereoscopic head-mounted displays. But the
reconstructed scene was rendered with a reduced polygon count
and lacked textures because of the technical limitations at that
time, which greatly influenced the 3D perception. Ren et al. [21]
presented another 3D modelling method using completed models
from an extensive crime scene model database. This technique has
limited reliability and accuracy, and the number of models is
insufficient. Furthermore, building a digital crime scene from a
model database is a time-consuming process. Our system records
the true proportions of the surroundings in the field by VR without
reducing the data during laser scanning, thus giving the secondary
survey personnel the most realistic site survey experience. With a
wireless connection, changing the position of the user in real-time
becomes easier. The function of performing measurements,
capturing screens and reading annotations in a virtual scene gives
the expert additional explanatory tools and represents a useful
approach for training police academy students.
Various scientific analyses of the patterns and locations of
physical evidence were performed. As Urschler [22,23] expected,
our work was extended to include more 3D data on the crime scene
and to show the scene through VR. In our case, the virtual crime
scene, external scans of the corpse, footprints, bloodstains, and
potential murder weapons were combined to help experts clarify
the course of the case. To our regret, we failed to obtain computed
tomography (CT) scans to document the victim’s internal injuries
due to legal restrictions. Ideally, the 3D surface data and CT data
could be studied at the same time to evaluate their spatial
interrelations [1,2,22,24].
In the BPA analysis, the region of origin and the trajectories
were shown in a precise virtual scene. Image recognition
techniques were used to reduce the workload of BPA experts.
However, it should be emphasized that BPA experts must be
consulted to eliminate erroneous results, and the centre of origin
was determined in an approximate sense. In our future work, the
influence of gravity and drag forces will be considered [25].
Three-dimensional optical surface scanning provides an alterna-
tive for documentation and can measure footprints or footwear in a
non-destructive way. With the 3D overlay method, one different
footprint among many can be differentiated and a bare footprint left
at the scene of the crime by the perpetrator can be identified. The
distribution of footprints reflects the movement of the murderer and
indicates the progression of the event. A 3D scene displayed by VR
provides considerable information on the spatial distribution of
footprints and makes it possible to trace the murderer's route.
The combination of 3D data and the VR system could improve
the comprehension and spatial memory of complex evidence for
courtroom audience or juries. However, similar to other new
technologies, the admissibility of the reconstructions as courtroom
evidence should be discussed because they come with the risk of
bias and abuse. The presentation includes objective results as well
as subjective hypotheses. Due to the high persuasiveness of this
technique, juries may be influenced by the associated visual aids. In
an attempt to reduce this bias, all digital evidence should be
carefully evaluated by experts. For combined information from
different sources, experts in every field are needed, such as police,
forensic pathologists, forensic podiatrists, BPA experts, etc. Every
hypothesis should be presented to the court when a dispute occurs.
Traditional evidence, such as medico legal evaluations and police
reports, still have a significant influence on the reconstruction and
should be presented together with computer-generated evidence.
In court, an expert witness must state the accuracy of both the
original data used to reconstruct the incident as well as the
reconstruction. The privacy of victim is another issue that should
be considered.
5. Conclusion
In the present study, we applied an integrated 3D scanning
platform designed for the crime scene documentation process to
obtain all the potential evidence in an entire crime scene and allow
an immersive and interactive investigation. All the processes can
be completed in a limited time, and the data can be transferred
between computers.
 J. Wang et al. / Forensic Science International 303 (2019) 109943
CRediT authorship contribution statement
Jinming Wang: Data curation, Formal analysis, Writing -
original draft, Investigation, Methodology. Zhengdong Li: Concep-
tualization, Formal analysis, Writing - original draft, Investigation,
Methodology. Wenhu Hu: Investigation, Software. Yu Shao: Data
curation, Formal analysis. Liyang Wang: Formal analysis, Investi-
gation. Rongqi Wu: Resources, Investigation. Kaijun Ma: Resour-
ces, Investigation. Donghua Zou: Funding acquisition, Project
administration, Supervision, Writing - review & editing. Yijiu
Chen: Funding acquisition, Project administration, Supervision,
Writing - review & editing.
Acknowledgements
This work was supported by the National Key Research and
Development Plan (No. 2016YFC0800702), Council of National
Science Foundation of China (Nos. 81701863, 81722027,
81571851), Shanghai Key Laboratory of Forensic Medicine (No.
17DZ2273200), Shanghai Forensic Service Platform (No.
19DZ2290900), Central Research Institute Public Project (No.
GY2018G-3) and Opening Project of Shanghai Key Laboratory of
Crime Scene Evidence (No. 2017XCWZK06).
References
[1] C. Villa, M.J. Flies, C. Jacobsen, Forensic 3D documentation of bodies: simple
and fast procedure for combining CT scanning with external photogrammetry
data, J. Forensic Radiol. Imaging 12 (2018) e2–e7, doi:http://dx.doi.org/
10.1016/j.jofri.2017.11.003.
[2] S. Kottner, L.C. Ebert, G. Ampanozi, M. Braun, M.J. Thali, D. Gascho, VirtoScan—a
mobile, low-cost photogrammetry setup for fast post-mortem 3D full-body
documentations in x-ray computed tomography and autopsy suites, Forensic Sci.
Med. Pathol.13 (1) (2017) 34–43, doi:http://dx.doi.org/10.1007/s12024-016-9837-2.
[3] P. Urbanova, P. Hejna, M. Jurda, Testing photogrammetry-based techniques for
three-dimensional surface documentation in forensic pathology, Forensic Sci.
Int. 250 (2015) 77–86, doi:http://dx.doi.org/10.1016/j.forsciint.2015.03.005.
[4] A. Shamata, T. Thompson, Using structured light three-dimensional surface
scanning on living individuals: key considerations and best practice for
forensic medicine, J. Forensic Leg. Med. 55 (2018) 58–64, doi:http://dx.doi.org/
10.1016/j.jflm.2018.02.017.
[5] A. Shamata, T. Thompson, Documentation and analysis of traumatic injuries in
clinical forensic medicine involving structured light three-dimensional
surface scanning versus photography, J. Forensic Leg. Med. 58 (2018) 93–
100, doi:http://dx.doi.org/10.1016/j.jflm.2018.05.004.
[6] S.A. Blackwell, R.V. Taylor, I. Gordon, C.L. Ogleby, T. Tanijiri, M. Yoshino, M.R.
Donald, J.G. Clement, 3-D imaging and quantitative comparison of human
dentitions and simulated bite marks, Int. J. Legal Med. 121 (1) (2007) 9–17, doi:
http://dx.doi.org/10.1007/s00414-005-0058-6.
[7] M.A. Bush, P.J. Bush, H.D. Sheets, Similarity and match rates of the human
dentition in three dimensions: relevance to bitemark analysis, Int. J. Legal Med.
125 (6) (2011) 779–784, doi:http://dx.doi.org/10.1007/s00414-010-0507-8.
[8] N. Lynnerup, M.L. Clausen, A.M. Kristoffersen, H. Steglich-Arnholm, Facial
recognition and laser surface scan: a pilot study, Forensic Sci. Med. Pathol. 5 (3)
(2009) 167–173, doi:http://dx.doi.org/10.1007/s12024-009-9094-8.
9
[9] J. Michon  ski, E.N. Hołowko, K. Lech, G. Ma˛czkowski, K. Januszkiewicz, R. Sitnik,
P. Bolewicki, M.M. Adamczyk, J. Michon  ski, E.N. Hołowko, Three-Dimensional
Measurement System for Crime Scene Documentation, Society of Photo-
Optical Instrumentation Engineers, 2017 p. 10.
[10] D. Raneri, Enhancing forensic investigation through the use of modern three-
dimensional (3D) imaging technologies for crime scene reconstruction, Aust. J.
Forensic Sci. 50 (6) (2018) 697–707, doi:http://dx.doi.org/10.1080/
00450618.2018.1424245.
[11] M.J. Bolliger, U. Buck, M.J. Thali, S.A. Bolliger, Reconstruction and 3D
visualisation based on objective real 3D based documentation, Forensic Sci.
Med. Pathol. 8 (3) (2012) 208–217, doi:http://dx.doi.org/10.1007/s12024-011-
9288-8.
[12] U. Buck, S. Naether, B. Räss, C. Jackowski, M.J. Thali, Accident or homicide –
virtual crime scene reconstruction using 3D methods, FSI 225 (1) (2013) 75–
84, doi:http://dx.doi.org/10.1016/j.forsciint.2012.05.015.
[13] L.C. Ebert, T.T. Nguyen, R. Breitbeck, M. Braun, M.J. Thali, S. Ross, The forensic
holodeck: an immersive display for forensic crime scene reconstructions,
Forensic Sci. Med. Pathol. 10 (4) (2014) 623–626, doi:http://dx.doi.org/
10.1007/s12024-014-9605-0.
[14] D.C. Niehorster, L. Li, M. Lappe, The accuracy and precision of position and
orientation tracking in the HTC vive virtual reality system for scientific
research, iPerception 8 (3) (2017), doi:http://dx.doi.org/10.1177/
2041669517708205.
[15] J.A.D. Denis Wesley Vernon, Forensic Podiatry: Principles and Methods, CRC
Press, Boca Raton, 2017, doi:http://dx.doi.org/10.1201/9781315395029.
[16] U. Buck, K. Busse, L. Campana, C. Schyma, Validation and evaluation of
measuring methods for the 3D documentation of external injuries in the field
of forensic medicine, Int. J. Legal Med. 132 (2) (2018) 551–561, doi:http://dx.
doi.org/10.1007/s00414-017-1756-6.
[17] Y. Chen, State of the art in post-mortem forensic imaging in China, Forensic
Sci. Res. 2 (2) (2017) 75–84, doi:http://dx.doi.org/10.1080/20961790.2017.
1337501.
[18] Z. Li, D. Zou, J. Zhang, Y. Shao, P. Huang, Y. Chen, Use of 3D reconstruction of
emergency and postoperative craniocerebral CT images to explore cranioce-
rebral trauma mechanism, Forensic Sci. Int. 255 (2015) 106–111, doi:http://dx.
doi.org/10.1016/j.forsciint.2015.07.012.
[19] A. Marcin, S. Maciej, S. Robert, W. Adam, Hierarchical, three-dimensional
measurement system for crime scene scanning, J. Forensic Sci. 62 (4) (2017)
889–899, doi:http://dx.doi.org/10.1111/1556-4029.13382.
[20] C. Reichherzer, A. Cunningham, J. Walsh, M. Kohler, M. Billinghurst, B.H.
Thomas, Narrative and spatial memory for jury viewings in a reconstructed
virtual environment, IEEE Trans. Vis. Comput. Graph. 24 (11) (2018) 2917–
2926, doi:http://dx.doi.org/10.1109/tvcg.2018.2868569.
[21] P. Ren, M. Zhou, J. Liu, Y. Fan, W. Zhao, W. Shui, Sketch-based modeling and
immersive display techniques for indoor crime scene presentation, in: P.
Matoušek, M. Schmiedecker (Eds.), Digital Forensics and Cyber Crime, Springer
International Publishing, Cham, 2018, pp. 181–194.
[22] A. Bornik, M. Urschler, D. Schmalstieg, H. Bischof, A. Krauskopf, T. Schwark, E.
Scheurer, K. Yen, Integrated computer-aided forensic case analysis, presenta-
tion, and documentation based on multimodal 3D data, Forensic Sci. Int. 287
(2018) 12–24, doi:http://dx.doi.org/10.1016/j.forsciint.2018.03.031.
[23] M. Urschler, A. Bornik, E. Scheurer, K. Yen, H. Bischof, D. Schmalstieg, Forensic-
case analysis: from 3D imaging to interactive visualization, IEEE
Comput. Graph. Appl. 32 (4) (2012) 79–87, doi:http://dx.doi.org/10.1109/
mcg.2012.75.
[24] S.A. Bolliger, M.J. Thali, Imaging and virtual autopsy: looking back and forward,
Philos. Trans. R. Soc. Lond. B: Biol. Sci. 370 (1674) (2015), doi:http://dx.doi.org/
10.1098/rstb.2014.0253.
[25] U. Buck, B. Kneubuehl, S. Näther, N. Albertini, L. Schmidt, M. Thali, 3D
bloodstain pattern analysis: ballistic reconstruction of the trajectories of blood
drops and determination of the centres of origin of the bloodstains, FSI 206 (1)
(2011) 22–28, doi:http://dx.doi.org/10.1016/j.forsciint.2010.06.010.