OPENCRIMESCENE REVIEW LOG

Interaction Log in a Virtual Crime Scene Investigation Learning Environment

Angela Brennecke, Stefan Schlechtweg and Thomas Strothotte

Department of Simulation and Graphics, Otto-von-Guericke University of Magdeburg, Germany

Keywords:

Virtual Learning Environments, Interaction Log, Visualisation.

Abstract:

In this paper we present a concept for visualising an interaction log in a virtual crime scene investigation

learning environment called OPENCRIMESCENE. The interaction log shall be used for reviewing the user’s

behaviour after having examined the virtual crime scene. Furthermore, it shall serve as a valuable discussion

background when it comes to self or teacher’s control. As initial point the log visualisation therefore has to

present an instant overview of user activity combined with signiﬁcant user interactions. In order to keep track

of the causal relations and the sequence of events we here focus on visualising time by combining different

signiﬁcant character positions in one image. This paper primarily aims at giving an overview of our system

and introduces the review log as well as its ﬁrst visualisation form.

1 INTRODUCTION

Due to the technological progress in Computer

Graphics in recent years, 3D environments offer the

possibility to lift various processes from real life

onto a virtual level. This ranges from simple 3D

model viewers to complex 3D learning environments

or computer games. Because of latest improve-

ments in 3D game engine’s development even non-

programmers can realise their own 3D virtual envi-

ronment. However, mapping real life procedures onto

a virtual level still is quite a challenge. This is also

because character animation usually cannot keep up

with real life motion.

The OPENCRIMESCENE project is a research

project on developing a 3D learning environment for

virtual crime scene investigation. It is conducted in

collaboration between the local University of the Po-

lice and our Computer Graphics Group. The project

primarily focusses on the realisation of procedures for

securing a crime scene. In this context we also pursue

realistic modelling of traces like ﬁnger- or shoeprints

as well as tools to secure those traces. The virtual

crime scene is implemented using the 3D game en-

gine Delta3D

and is currently in a prototypical state.

The OPENCRIMESCENE system is divided into

three parts: First, the authoring mode where the

teacher can set up a virtual crime scene, arrange

traces, and prepare for, e. g., a virtual examination.

Second, a practice mode in which the student can re-

capitulate what he or she has learned in practical ex-

perience. This will also serve as examination mode.

Third, we want the system to see over the virtual

crime scene and allow for analysing the student’s be-

haviour in a review mode.

Seen from the learning environment point of view

the central question is: How can real life procedures

of crime scene investigation be turned into virtual in-

teraction techniques which can be logged for review-

ing the learning development of the student? The fo-

cus therefore does not only lie on the realistic simula-

tion of the crime scene investigation but also on the

visualisation of the virtual interactions themselves.

Hence, we want to make it possible to actually see

what the student has done in a visual log – the OPEN-

CRIMESCENE review log.

http://www.delta3d.org

185

Brennecke A., Schlechtweg S. and Strothotte T. (2007).

OPENCRIMESCENE REVIEW LOG - Interaction Log in a Virtual Crime Scene Investigation Learning Environment.

In Proceedings of the Second International Conference on Computer Graphics Theory and Applications - AS/IE, pages 185-190

DOI: 10.5220/0002079501850190

 SciTePress

In this paper we introduce our concept for this log.

It aims at giving an instant overview of the student’s

behaviour when securing the crime scene. Therefore,

signiﬁcant interactions have to be captured. These are

then to be displayed by different visualisation forms.

Starting from the log’s global view there will also be

the possibility to take a closer look at certain events

either by scene replays or more detailed logs.

Our ﬁrst visualisation form that is presented here

combines signiﬁcant character positions in one image.

We concentrate on encoding action over time by ap-

plying different rendering techniques. Even though

we focussed on small scenes with only one viewing

angle, we receive a ﬁrst impression of user movement

and behaviour.

2 THE REVIEW LOG

One main advantage of a virtual learning environ-

ment is that even complex scenarios like, e.g. a crime

scene, can be represented virtually and, thus, be expe-

rienced even by multiple users whenever they want.

Additionally, users can beneﬁt from the fact that in-

teractions within a virtual environmentcan be logged.

Thus, they can step back and take over a more ob-

jective perspective on how his or her behaviour has

been. Such an interaction log can also be helpful for

teacher’s control, not only in terms of a virtual exam-

ination but also for discussion.

The fact that virtual environments support syn-

chronous as well as asynchronous communication is

crucial when it comes to learning and understanding.

Here, we focus on asynchronous communication ﬁrst

in the form of the visual review log. We have two rea-

sons for that: First, we want the system to be quickly

useable and have therefore postponed the realisation

of multi-user access. Second, we believe that virtual

environments cannot reproduce crime scene investi-

gation procedures in such a detail that would hold for

real life techniques. Hence, an afterwards discussion

is still essential with the review log as initial point.

What is required for such an interaction log?

Speaking in terms of crime scene investigation, the

securing of traces is one of the major tasks to be ful-

ﬁlled. There are certain rules a police student has

to internalise. How to enter the crime scene without

damaging existing traces? How to secure the traces

with the correct tools, for instance only certain pow-

ders are best for securing ﬁngerprints? Besides, there

are crucial regions to look for traces ﬁrst, e. g. ﬁnger-

prints will rather be found at doorknobs than on the

ﬂoor, and so on.

As a result the review log has to visualise impor-

tant interactions, user awareness as well as false or

missing actions and behaviours. However, at the cur-

rent stage of implementation we cannot decide auto-

matically for signiﬁcance of interactions and user be-

haviour. We rather take a look at how to combine dif-

ferent character positions in one image in a pleasing

way. Hence, as a starting point for the review log we

simply track the character’s position and posture over

time. The combination in one image leads to a ﬁrst

impression of an instant user behaviour overview.

3 RELATED WORK

3D virtual environments have become a major appli-

cation area for research. They not only offer graph-

ically pleasing visualisations for complex scenarios

but also the possibility to explore them in single or

in multi-user mode.

Beside collaborative work and entertainment,

learning is one of the most explored application ar-

eas of virtual environments (Mondjar-Andreu et al.,

2006). Several different approaches towards de-

signing such environments have been introduced,

e.g. (de Antonio et al., 2005; Bouras et al., 2002;

de Oliveira et al., 2000). Likewise, the application

variety ranges from virtual hairstyling (Ward et al.,

2006) to maths classes (Elliott and Bruckman, 2002)

and many more.

Although most environments allow for syn-

chronous online as well as asynchronous ofﬂine ex-

perience, realising collaboration is still a challenging

task. As a result, innovative system frameworks of-

ten remain prototypes, e. g. (Lombardi and Lombardi,

2005). Thus, we focus on asynchronous communica-

tion ﬁrst. This will be achievedby logging relevant in-

teraction sequences in order to communicate student

behaviour for review.

Current works on recording, summarising and vi-

sualising relevant user interaction are few. Neverthe-

less replay functionality is a required and common

feature in most computer games and virtual environ-

ments, e. g. (Wagner, 2004; Logan et al., 2002). The

challenge, however, is to decide for importance with-

out losing causal relationships.

(Halper and Masuch, 2003) addressed the problem

of summarising only signiﬁcant game events. Their

approach was to analyse game state variables and to

visualise these as a series of according game snap-

shots afterwards. Although the attempt is promising,

further improvements could be incorporated, e. g.,

different views of the scene to allow for a better con-

nection of game event relationships. (Friedman et al.,

2004) generate movie summaries from virtual envi-

GRAPP 2007 - International Conference on Computer Graphics Theory and Applications

186

ronment logs. They put emphasis on causal relations

to support story understanding.

In contrast, the works of (Chittaro and Ieronutti,

2004; Hoobler et al., 2004) concentrate on overall

user activities by highlighting user navigation paths.

Furthermore, Hoobler et al. also encode action over

time by colouring whole game environment areas.

Yet, the resulting overview images often take into ac-

count current activities only.

(Fielding et al., 2006) use reporter, editor and pre-

senter agents to log, ﬁlter and visualise important

events in a game environment. Their approach’s nov-

elty lies in the embodied reporter agents which try to

balance player’s and spectator’s needs. Yet, no pre-

senter agent has been realised entirely.

Although not focussing on summarising events,

(Grammenos et al., 2006) have studied user activity

in virtual environments in depth. They introduce the

Virtual Prints as a concept to track user navigation,

trace user interaction and leave marks to communi-

cate to other users. Thus, beside game state variables

also virtual prints could be used to make assumptions

on the signiﬁcance of user interactions.

Finally, speaking of virtual crime scene investi-

gation itself most research activities focus on static

photo-realistic crime scene reconstruction (Se and Ja-

siobedzki, 2005; Gibson and Howard, 2000; Howard

et al., 2000). There is only the CRIME SCENE CRE-

ATOR by (Davies et al., 2004) which addresses simi-

lar issues as we do. Although the system is also in-

tended for crime scene reconstruction only animated

characters can be added to the created scenes. These

can then act out simple crimes. Moreover, variable

camera views make it possible to interactively view

the scenes from different angles. The project’s main

focus is to support investigators by allowing for a

fast and efﬁcient evaluation of crime scenes and se-

quences of events. In contrast to our system, how-

ever, neither interactive crime scene investigation nor

teaching purposes are considered.

4 LOG VISUALISATION

As stated above, logging the student’s behaviour

in the learning environment is the basis for asyn-

chronously communicating how the crime scene in-

vestigation has taken place and which actions the stu-

dent has done. However, our system is currently at

a development stage which does not allow for auto-

matic event logging yet. As a consequence we record

signiﬁcant events manually in order to concentrate on

the log’s visualisation.

Having examined the works from section 3 the vi-

sualisation of activities in games or virtual environ-

ments can generally be divided into two groups, taken

animated replays aside. First, important events are

presented in a series of snapshots, and second, regions

with strong user presence are being highlighted. We

call the former an event-based or local visualisation

form putting emphasis on single events and the latter

a course-based or global visualisation form represent-

ing overall activities.

In the OPENCRIMESCENE system we want to

achieve a visualisation form which combines both.

We need an overview of global user activity as well as

the depiction of signiﬁcant local interactions in order

to receive an instant user behaviour summary. As a

starting point we decide to use a set of snapshots from

the character in the virtual world and combine the

character’s positions over time in one image. To facil-

itate the combination process further we use one static

camera to record the character movement. Moreover

we assume that only non-overlapping positions have

been found.

4.1 Visualising Time

Simply combining certain character positions does

not make any assumptions about the global process of

crime scene investigation. Therefore, we need to en-

code time. As we only use one static camera position

the only thing changing in the virtual environment is

the virtual character. Hence, a possible solution to

encode time would be to change the character’s ren-

dering in various ways.

Transparency: Probably the most natural way to

encode time is to make the character more transpar-

ent as time passes. Hence, in the ﬁrst stages in a crime

scene investigation, the character is visualised highly

transparentand the later an action is depicted the more

opaque the character becomes, see Figure 1.

This visualisation is controlled by a linear func-

tion that maps the time to the level of transparency,

however, this function needs to be designed in such a

way that the transparency value of the ﬁrst position to

be visualised is not 100% in order to ensure the visi-

bility of all positions.

Furthermore, it is important to make all posi-

tions distinguishable, therefore, the sole use of trans-

parency is only possible, if two points in time are not

too close together, see Figure 2.

However, if the character positions are distin-

guishable but user movement has been fast, trans-

parency is also limited in indicating time differences,

as shown in Figure 3. In those cases, additional visual

clues have to be used. One solution is to employ any

OPENCRIMESCENE REVIEW LOG - Interaction Log in a Virtual Crime Scene Investigation Learning Environment

187

Figure 1: Encoding position over time by different trans-

parency values.

Figure 2: Transparency limits due to an accumulation of

close character positions, see positions 1 and 2.

Figure 3: Transparency limits due to fast user movement.

The sequence of positions 1 and 4 is hardly recognisable

whereas positions 2 and 3 can be well distinguished.

user-deﬁned mapping function to individually deﬁne

transparency values for the single positions.

Colours: Another way of encoding time is to use

colours on a colour scale and to map points in time to

which the respective character position belongs to the

chosen colour values. To do so, the image of the char-

acter is ﬁrst de-saturated to achieve a grey value ver-

sion and then re-coloured in the respective colour. If

the starting point of this re-colouring is a more or less

transparent version of the character image, the trans-

parency clue can be enhanced. Otherwise the used

colour scale is the only way in which time is visu-

alised. This is somewhat problematic, since colour

scales that are built on the basis of changing hues are

barely useable when it comes to the visualisation of

ordinal values (as time is in our case). Therefore,

colour scales that are built on one speciﬁc hue with

changing intensity are a better tool when no trans-

parency is employed.

Silhouettes: One major concern when superimpos-

ing multiple positions of the character in the same

image is that the character might not be optimally

visible in a rather cluttered image showing multiple

characters and the objects being present at the crime

scene. Drawing the silhouettes of the character makes

it stand out against the background of the crime scene

and, hence, better visible, see Figure 4. The visual de-

Figure 4: Using silhouettes to further enhance the charac-

ter’s visibility.

sign of these silhouettes can be adapted to the visual-

isation of the character, i. e., their colour can be set to

ﬁt the colour of the character. On the other hand, the

silhouette style can also be used to map different val-

ues, like importance or duration of time, see Figure 5.

However, the priority in the visualisation is always to

be set in such a way that the order of events becomes

clearly visible.

4.2 Combining the Techniques

The rendering techniques presented above show dif-

ferent renderings to encode time. We believe that

transparency is the main indicator for time-sensitivity

in this scenario. Yet, transparency is limited when

GRAPP 2007 - International Conference on Computer Graphics Theory and Applications

188

Figure 5: Using coloured silhouettes to additionally encode

time. For instance, fast user movement is indicated by the

blue silhouettes.

it comes to visualising events which are very close

in time. Silhouettes could additionally be applied ei-

ther to better differentiate between close positions or

to directly encode time-lags. However, the latter does

not hold for short changes in time. Figure 6 shows

a combination of both techniques with an additional

legend relating transparency and silhouette colours to

the actual time values. This seems to be an adequate

visualisation for a small number of signiﬁcant events.

Figure 6: Combining the enhancement techniques in one

image.

5 CONCLUSION

Our ﬁrst visual review log shows how to combine dif-

ferent snapshots of character positions in one image.

Furthermore, mapping rendering style parameters to

time values seems to be an appropriate method to

track the order of user activity. In this ﬁrst scenario

transparency seems to be the strongest indicator for

time-lags whereas silhouettes increase the perception

of the single character positions. Yet, the presented

visualisation form concentrates on showing temporal

relationships and signiﬁcant character positions in a

small scene only — without taking into account dif-

ferent camera angles, problems of cluttered images

and the like. We nevertheless receive a ﬁrst impres-

sion of how to combine a local and a global visualisa-

tion form and of how an instant user behaviour sum-

mary may look like.

6 FUTURE WORK

Having introduced OPENCRIMESCENE as well as the

review log concept with a ﬁrst visualisation form we

now would like to address some of our ideas for future

research. A ﬁrst task will be to implement methods

for automatic event logging and visualisation. Among

others, this will also imply the integration of multiple

camera views into the scene in order to decide for the

event’s best view.

Besides, we will need to further investigate the

ﬁeld of information visualisation, e. g. (Ware, 2000).

The event’s size or the silhouette’s thickness might

be better indicators for action over time than colour

and transparency are. Colour could instead be used

for grouping related interactions. A user study will

therefore be essential to examine the rendering styles’

applicability.

Above, we would like to realise additional vi-

sualisation forms, e.g. comic-like logs. Combin-

ing the works of (Halper and Masuch, 2003) and

(Friedman et al., 2004) may serve as a starting point

here. For this purpose rendering techniques like non-

photorealistic or multi-perspective rendering shall be

explored, e. g. (Strothotte and Schlechtweg, 2002;

Glassner, 2000). The former to convey more and dif-

ferent information and the latter to describe causal re-

lationships.

ACKNOWLEDGEMENTS

We would like to thank our colleagues from the Uni-

versity of the Police for their support and knowledge

sharing. Marie-Luise Mueller for her works on the

review log’s ﬁrst visualisation form (Mueller, 2006).

All pictures shown here are taken courtesy of hers.

OPENCRIMESCENE REVIEW LOG - Interaction Log in a Virtual Crime Scene Investigation Learning Environment

189

REFERENCES

Bouras, C., Triantaﬁllou, V., and Tsiatsos, T. (2002). A

Framework for Intelligent Virtual Training Environ-

ment: The Steps from Speciﬁcation to Design. Edu-

cational Technology & Society, 5(4):11–26.

Chittaro, L. and Ieronutti, L. (2004). A Visual Tool for

Tracing Users’ Behavior in Virtual Environments. In

Proceedings of the Working Conference on Advanced

Visual Interfaces 2004, pages 40–47, New York, NY,

USA. ACM Press.

Davies, N., Mehdi, Q., and Gough, N. (2004). Virtual Crime

Scene Reconstruction with Integrated Animated Char-

acters. In Proceedings of the 18

European Simu-

lation Multiconference, pages 338–393, Magdeburg,

Germany. ECMS.

de Antonio, A., Ramirez, J., Imbert, R., and Mendez, G.

(2005). Intelligent Virtual Environments for Train-

ing: An Agent-Based Approach. In 4th International

Central and Eastern European Conference on Multi-

Agent Systems, CEEMAS 2005, volume 3690 of Lec-

ture Notes in Computer Science, pages 82–91, Berlin

Heidelberg. Springer.

de Oliveira, J. C., Shirmohammadi, S., Hosseini, M.,

Cordea, M., Georganas, N., Petriu, E., and Petriu,

D. (2000). Virtual Theater for Industrial Training:

A Collaborative Virtual Environment. In 4

Interna-

tional Multiconference on Circuits, Systems, Commu-

nications & Computers, pages 294–299, Vouliagmeni,

Athens, Greece. WSEAS.

Elliott, J. and Bruckman, A. (2002). Design of a 3D Inter-

active Math Learning Environment. In Proceedings

of the Conference on Designing Interactive Systems,

pages 64–74, New York, NY, USA. ACM Press.

Fielding, D., Logan, B., and Benford, S. (2006). Balancing

the Needs of Players and Spectators in Agent-based

Commentary Systems. In 5

International Joint Con-

ference on Autonomous Agents and Multiagent Sys-

tems, Hakodate, Japan, May 8-12, 2006, pages 996–

998, New York, NY, USA. ACM Press.

Friedman, D. A., Feldman, Y. A., Shamir, A., and Dagan,

T. (2004). Automated Creation of Movie Summaries

in Interactive Virtual Environments. In Proceedings

of the IEEE Virtual Reality Conference 2004, pages

191– 200, Los Alamitos, CA, USA. IEEE Computer

Society.

Gibson, S. and Howard, T. (2000). Interactive Reconstruc-

tion of Virtual Environments from Photographs, with

Application to Scene-of-Crime Analysis. In Proceed-

ings of the ACM Symposium on Virtual Reality Soft-

ware and Technology, pages 41–48, New York, NY,

USA. ACM Press.

Glassner, A. (2000). Cubism and Cameras: Free-form Op-

tics for Computer Graphics. Technical report, Mi-

crosoft Research MSR-TR-2000-05.

Grammenos, D., Mourouzis, A., and Stephanidis, C.

(2006). Virtual Prints: Augmenting Virtual Environ-

ments with Interactive Personal Marks. International

Journal of Human-Computer Studies, 64(3):221–239.

Halper, N. and Masuch, M. (2003). Action Summary for

Computer Games – Extracting and Capturing Action

for Spectator Modes and Summaries. In Proceedings

of 2nd International Conference on Application and

Development of Computer Games, pages 124–132,

Hong Kong, China.

Hoobler, N., Humphreys, G., and Agrawala, M. (2004).

Visualizing Competitive Behaviors in Multi-User Vir-

tual Environments. In Proceedings of the Conference

on Visualization 2004, pages 163–170, Los Alamitos,

CA, USA. IEEE Computer Society.

Howard, T. L. J., Murta, A. D., and Gibson, S. (2000). Vir-

tual Environments for Scene of Crime Reconstruction

and Analysis. In Proceedings of SPIE: Visual Data

Exploration and Analysis VII, volume 3960, pages

41–48, Bellingham WA, USA. Society of Photo-

Optical Instrumentation Engineers.

Logan, B., Fraser, M., Fielding, D., Benford, S., Green-

halgh, C., and Herrero, P. (2002). Keeping in Touch:

Agents Reporting from Collaborative Virtual Environ-

ments. In Artiﬁcial Intelligence and Interactive En-

tertainment: Papers from the 2002 AAAI Symposium,

pages 62–68, Menlo Park, CA. AAAI Press. Techni-

cal Report SS–02–01.

Lombardi, M. M. and Lombardi, J. (2005). Croquet Learn-

ing Environments: Extending the Value of Campus

Life into the Online Experience. 3

International

Conference on Creating, Connecting, and Collaborat-

ing through Computing, 00:135–142.

Mondjar-Andreu, R., Garcia-Lopez, P., Pairot-Gavalda,

C., and Gmez-Skarmeta, A. F. (2006). Track-

ing the Evolution of Collaborative Virtual Environ-

ments. UPGRADE - The European Online Mag-

azine for the IT Professional, http://www.upgrade-

cepis.org, VII(2):25–30.

Mueller, M.-L. (2006). Benutzerschnittstelle fr die Tatort-

sicherung. Master’s thesis, Department of Simula-

tion and Graphics, Otto-von-Guericke University of

Magdeburg.

Se, S. and Jasiobedzki, P. (2005). Instant Scene Modeler

for Crime Scene Reconstruction. In Proceedings of

the 2005 IEEE Computer Society Conference on Com-

puter Vision and Pattern Recognition - Workshops,

page 123, Los Alamitos, CA, USA. IEEE Computer

Society.

Strothotte, T. and Schlechtweg, S. (2002). Non-

Photorealistic Computer Graphics. Modeling, Ren-

dering, and Animation. Morgan Kaufmann Publish-

ers, San Francisco, CA, USA.

Wagner, C. (2004). Developing Your Own Re-

play System. Gamasutra Online Article,

http://www.gamasutra.com/features/20040204/

wagner 01.shtml.

Ward, K., Galoppo, N., and Lin, M. C. (2006).

A Simulation-based VR System for Interactive

Hairstyling. In Proceedings of the IEEE Virtual Re-

ality Conference 2006, volume 0, pages 257–260, Los

Alamitos, CA, USA. IEEE Computer Society.

Ware, C. (2000). Information Visualization. Academic

Press, San Diego, 2

edition.

GRAPP 2007 - International Conference on Computer Graphics Theory and Applications

190