Electronic administration software, GIS and a 3D scanner
as enhancement for the crime scene investigator
36 EC February 2015 August 2015 Hüsnü Tas 5975425 Msc in Forensic science, University of Amsterdam Examiner: Zeno Geradts, Supervisor: Jurrien Bijhold Internship location: Netherlands Forensics Insitute 2015 Suggested Journal: Forensic Science InternationalPreface
I would like to thank Jurrien Bijhold and Zeno Geradts for their help and guidance throughout the research process. I would also like to thank all members of the Mobile Forensic Team for their wilingness in answering my questions. Furthermore I would like to thank Reinoud Stoel for his help with the statistical part of the study. I would also like to thank the forensic investigators from Rotterdam, Den Haag and Amsterdam for their cooperation in this study. Especially Paul and Karin from Rotterdam helped me a lot by registering the mock crime scene. And of course I would like to thank all NFI employees that volunteered in my experiments. Without them, this study would not be possible!Contents
Preface 2 Introduction 4 Materials 6 PDadmin 6 DotProduct 3D scanner 7 NFI CSI lab 9 Interviews 10 Method 10 Results 10 Literature study 17 Method 17 Results 17 User tests 27 Method 27 Results 32 Conclusion and discussion 39 Future work 41 References 42 Appendix A: Police burglary case envelope 45 Appendix B: Instructions 47 Appendix C: Evaluation form 48 Appendix D: iFEA 49 Appendix E: Method 1: Paper tracelist 50 Appendix F: Method 2: PDadmin 51 Appendix G: Method 3: Panorama 52 Appendix H: Method 4: 3D scan 53Introduction
Problem statement Currently, crime scene investigators (CSI) record the traces found on a crime scene (CS) on a piece of paper or in a digital file in the form of a word or excel file. A major part of the report is written afterwards. When the CSI is securing a trace, a photo is taken of the trace while holding a barcode with a SIN number in front of it. The photos are usually taken from a closeup perspective, which does not clearly provide the viewer enough information about the relative position of the trace and an overview of the crime scene. Besides a closeup photo, a medium shot and a distance overview photo is usually taken for overview purposes. But this is not always easy to do, so the NFI wants to improve the current workflow. The traces and observations on a crime scene should always be well documented in a digital application. Another problem is that it is difficult to mark and visualize the location of latent traces. This needs to be done on the spot. A sticker does not seem to be able to fulfill this task. So to summarize the problem: The quality and efficiency of the chain of custody and also the overview of the CS should be improved. There have been several attempts in the past by the NFI to solve the above problems with different technologies. One of the attempts was with a case administration tool named PDadmin. PDadmin has been created as a possible solution to ease the work of a CSI by providing a framework where traces can be stored in a digital list and other information can be recorded. PDadmin is also meant to link photos with traces on the scene. Testing of the application has been done in a previous study. In our study, we have programmed some improvements in the application before we conducted more tests. This allowed us to find out if how feasible it was to improve the application in the future. Crime scene investigators are also increasingly interested in new ways of visualizing complex crime scenes. These can be Geographical Information Systems (GIS) and 3dimensional (3D) recording. We will use PDadmin as an example of a case management tool and the DotProduct 3D scanner as an example of a handheld scanner. In the case of GIS, we will look at the literature how GIS can aid the crime scene investigator. In the literature study, we will also look at related case administration and 3D tools.Research questions
Our main question is: How can the use of PDadmin, 3D scanning and GIS enhance the work of the Crime Scene Investigator? We will try to answer the main question by the formulation of the following subquestions: ● What can be improved about the current way of crime scene registration? ● What is the effect of the use of a 3D handheld scanner for a Crime Scene Investigation? ● What is the effect of the use of the PDadmin software compared to the paper administration? ● What is the effect of the use of GIS for a Crime Scene Investigator? ● What is the best way to combine PDadmin, GIS and 3D methods? ● How can the changes be validated? In this study, the following three research methods have been used to answer the above questions: interviews, a literature study and user experiments. These methods will be further explained in their dedicated chapters under the subchapter ‘Methods’. In the subsequent subchapter ‘Results’, the results of that research method are presented. In the chapter ‘Materials’, we will explain the new registration methods PDadmin and DotProduct which we have used in our user experiments. In the chapter ‘Conclusion and Discussion’, we will give an overview of our possible answers on our research questions and discuss how the recommendations can be implemented. In the chapter ‘Future work’, we will give advice on future studies and developments.Materials
PD-admin
An attempt to modernize the CSI was made by developing an application called ‘PD admin’. PDadmin is a program that is developed by three forensic ICT students during their internship at the NFI. The chain of custody helptool PDadmin is still in the beta phase and not used in case work yet. The reason for this is that the system still has some shortcomings and its advantages have not been proven yet. There are a lot of improvements that can be made for PDadmin. First of all, the software needs to be able to work completely standalone and has to connect to the NFI network later for data storage. At the moment, the wireless connection between the program and the camera is through an unencrypted connection. The encryption will be done by a third party, otherwise the program cannot be used securely at the crime scene. In PDadmin, pictures need to have a story as soon as possible. The PDadmin program should also be used to plan the investigation. Besides a barcode scanner to scan the SIN stickers, PDadmin should be able to export trace information that can be printed on 10x10cm stickers from a Zebra printer on the crime scene. This stickers can then be pasted on the envelope containing the trace. A SIN (Sporen Identificatie Nummer) is a unique number that is assigned to every individual trace found on a crime scene. A SIN sticker is always accompanied by a barcode, so it can be scanned into a computer. This reduces the chance of entering a wrong SIN on the computer. The bigger SIN stickers have a RFID chip that allows the tracking of the physical location of the trace. A SIN exists of four letters and digits and ends with NL. Features of PDadmin (image 1): ● Orientation: The CSI can write down notes and play recorded audio ● Case information: General information about the case, CSI on scene, the actions (searched for trace x with tool x) and tests are recorded ● Tracelist: Traces with SIN, date, time, photo and description are stored here ● Agreements: agreements with date and persons ● Export: Function to export any type of information in a Word file ● Apps: shortcuts to a 3D viewer, GIS application and a photo annotaton program ● XML editor: external application to edit CSI and techniques The PDAdmin software has been tested by a student already [Schroder]. The results of this study are pending. We will use a slightly modified version in our experiments.Image 1: PDadmin interface
DotProduct 3D scanner
The 3D scanner that we used to record our mock crime scene, is a handheld 3D scanner called Dot Product [Dotproduct] (image 2). This scanner may be ordered by the NFI in the near future. The DotProduct scanner consist of an Android tablet with an 8inch screen combined with a sensor and camera from Primesense. The tablet is an Nvidia tablet instead of a generic consumer type tablet. This tablet has a powerful graphical processing unit that makes it possible to smoothly view 3D images. Primesense is best known for their sensors used in the Microsoft Kinect. The scanner can export the data in PTS, PTX and PLY point cloud formats. Additionally, the data can be exported in a highly compressed format called DP. This format can be read by a number of 3D viewers. DotProduct comes in two types. A long range and close distance scanner. The long distance scanner is for 0.6 to 3.7 meter and the short distance is for 0.3 to 2 meter. The long distance scanner is used for a quick overview of the scene. The short distance scanner is used to record the scene in more detail. Dot product is not accurate for small objects, but good for general overview. DotProduct Usage The sensor has to be connected to the tablet first by microUSB. Then the tablet can be started. A default Android home screen can be seen. When the Phi.3D software is started, the cameraneeds to warm up for a minute before it can be properly used. The user can now scan the environment. This will take a few minutes. It is possible to use April tags to provide correspondence constraints to help with loop closure on large scenes. These are ordered by the Rotterdam police, but currently they are not using them. After the scan, the user can immediatly view the 3D file on the tablet. Before exporting the file, the scan has to be optimized first. This takes about thirty seconds. After that, an export is possible, this also takes around thirty seconds. It is important to set the top view (Z) in the program and selecting ‘user defined’ in the export settings, otherwise it will export the file with a random top view. This causes issues when viewing the file in Recap. When the proper top is determined, pressing the top button in Recap will show the scene from above. Besides an export of the 3D file, the image frames can also be exported. When you want to view the 3D file on a 3D viewer on a computer, you need to connect the tablet with the computer first. The fastest way to do this is to use a USB cable. You then need to copy the .dp file to the computer. This file is usually between 20 and 30 mb due to its compressed nature. Therefore the copy process only takes a few seconds. After this, the file can be imported in the 3D viewer, in our case Autodesk Recap. The import process takes a few minutes. An alternative to Autodesk Recap is CloudCompare. This program is used by the Rotterdam pd and it is free to use, but less user friendly than Autodesk Recap. Recap is free to use for three years for students, but normally it costs around 500€ for a years license. There is also a trial period of thirty days. Since march 2015, Arena 4D accepts DotProduct files, but with the free version only viewing is possible. A commercial license costs €1999. Image 2: DotProduct scanner
DotProduct can also export the images it used to create the 3D model. These are around hundred .bmp images with a resolution of 320 x 240. The size of one image is 226 kb. Altough the resolution of the images is low, it is more detailed compared to the 3D model. So the investigator can refer to these images if a closer look of an area is required. In this case, the 3D scanner acts as a photo camera. Currently, the photos have to be manually copied to the computer, but it should also be possible to copy them through WiFi just like it is possible with a photo camera since the tablet has WiFi too. In ReCap, the annotations can be accompanied with a title, a description and one or more images. In the annotation list, it is possible to disable the visibility of every single annotation. It is also possible to directly point the camera to a location by clicking on the annotation, but the viewpoint is always static, so the view may be obstructed by other objects, like a wall. [recap]
NFI CSI lab
The CSI lab (image 3) is located in the fieldlab, which is a building next to the NFI in the Hague. The CSI lab is used for training and experiments by the NFI. We will use the CSI lab as a mock crime scene. To test the registration techniques in a realistic setting, a complex crime scene should be created with an average amount of traces. For this test, the CSI lab at the NFI could be used. The CSI lab consists of an indoor house, street, interactive wall and an observation + briefing room. This location is used by the NFI as a mock crime scene location. So it is a very suitable place to test our techniques. The house has a bedrood, bathroom, hallway and a kitchen/living room.
Interviews
Method
Before we can answer the question ‘What can be improved about the current way of crime scene registration?’, it is neccessary to know how the crime scene investigators in the Netherlands do their work at this moment. To obtain this information, forensic investigation teams from different police regions and the NFI have been observed or interviewed. We have also demonstrated PDadmin to them and asked them for their feedback. Questions Some key questions were prepared before the observation of a CSI. The questions are mostly open questions to encourage the CSI to tell as much as possible. After the interviewees talked about their way of work, they were shown the PDadmin application and were asked to give feedback. Question 1: Can you describe the process of a crime scene investigation? Question 2: What tools (pen, paper, digital, panorama, 3d) do you use? Question 3: What roles are present on the CS and who is responsible for the report? Is it predetermined? Question 4: What frustrates you in the current case administration? Question 5: What do you miss in the current situation? Question 6: How do you look up to the idea of using technology in CSI? Question 7: How many traces do you find/collect on an average crime scene? What kind of traces are expected to be found? Question 8: What kind of cases do you do? Question 9: Which functionalities of PDAdmin are useful for you and which one are not? Question 10: What is your opinion/feedback on PDadmin? We have asked investigators from various police departments with crime scene experience if they wanted to participate in the interviews. In Rotterdam, we have observed and interviewed two forensic assistants and interviewed two investigators who had experience with complex crime scenes. In the Hague, we interviewed a senior forensic investigator and in Amsterdam, we interviewed three forensic investigators. A summary of these interviews can be seen below.Results
Forensic assistants Rotterdam police Two days were spent in the field with this team of crime scene investigators of the Rotterdam police. These people are called forensic assistants. They are not actual police officers, butBOAs (special investigating officers). They usually investigate (attempted) burglary cases and destruction of property after the first responders request CSI assistance to perform forensic investigation. The forensic assistants can do multiple cases a day and when it is busy, they do multiple cases right after each other. The burglary team consists of one person for simple cases during the day and of two people at night and more complex cases like weed grow rooms. They usually do not take photos at burglary cases. In the case of weed grow rooms, photos are always taken, but mostly for overview purposes. No SIN numbers are assigned on the crime scene due to time constraints. Most of the time, two techniques are used: the powdering technique on poreus surfaces to find fingermarks and the use of SIL for toolmark analysis. The time spent on a burglary case is roughly 45 minutes. There are usually few traces found. In most cases there are no fingermarks from the perpetrator present due to glove usage by burglars. When a trace is found, it is placed inside a gripbag. When the CSI returns to the police department, SIN numbers are assigned to the traces. The investigator writes all case information on one envelope (see Appendix A). The traces, which are sealed, are also put in that envelope. When the investigator is back at the bureau, the case information is entered into BVH (Basis Voorziening Handhaving). From BVH, it can be automatically imported into LSV (Landelijke Sporen Volgsysteem). Improvements current workflow PDAdmin was also used to record the traces on an attempted burglary case in Rotterdam, but it turned out that PDadmin is not suitable for burglaries, since it requires different information input as it can be seen in Appendix A. There are also much less traces found on a burglary case compared to a homicide case. The complaints at the police department are mostly related to the fact that they have to write down everything three times. Photos can not be directly linked to traces in BVH, they are stored on an internal network drive. The burglary team does not prefer to use a laptop. They already have to carry a case with forensic tools and sometimes a case with a photo camera. A tablet, which is smaller, is preferred. BVH BVH is an incident registration system used by all police forces in the Netherlands. Police reports are also stored here. BVH can be used to search for tags like addresses or license plates. There have been complaints about the quality of BVH for years. The system is not user friendly and causes the user to enter data multiple times. And although BVH should be a system where all the police information is merged from the whole country, that is not the case at this moment. The Rotterdam police can only look at cases from the Rotterdam area at this moment.
An application like PDadmin would not have an additive value at this moment for the burglary and homicide team, because they will have to retype everything in BVH. BVH does not support import from an external file, document or database. Everything has to be entered manually. Homicide team Rotterdam police There are also investigators for more serious cases, like homicides or suspected homicides. These people are called ‘1e lijns’, which means first liners. The first liners use trace envelopes and their own notebooks. No standard form is used when writing notes. Occasionally, at complex crime scenes, laptops are also used. Sometimes they assign a SIN number on the crime scene, but it is not a standard procedure. They mainly rely on photos and its metadata like time. When taking photos, they use trace numbers for the chain of custody. Sometimes the SIN numbers are not visible on photos. This is the case with photos from a large distance, for example in the case of an outside shooting. They solve this by writing the SIN number on the back of the sticker and placing it on top of a trace number, just like a restaurant menu card. When needed, panoramascanners and 3D scanners are used. They have total stations with GPS in them, photos and GIS is used together to create a map of the crime scene. The Rotterdam police department uses GISweb. This can be accessed by gis.rotterdam.nl. The website is very slow. It can be only used by Rotterdam police, but they can also not see maps of other locations. The amount of persons on a crime scene can differ. This is usually between one and four people and is highly dependant on the crime scene. When it is a crime scene on a public place or there is a time constraint for some reason, more people are needed to process the scene faster. At a complex crime scene, there is always a coordinator. When the victim is heavy, more people are needed to lift the body. There is someone who makes the photos and this is also the person that writes the chain of custody. At simple cases, for example when someone shot himself in the head, there are few traces and there is not much to describe. At complex crime scenes, there are obviously more traces. Instead of writing notes, they mostly use photos as a notebook. The timestamps, photos and numbers is sufficient for them. It is also possible to use the iPad as a notebook. All the cases are visible in the internal network of the forensic department of the Rotterdam police. At complex crime scenes, someone can be outside (in a tent) with a laptop to keep an administration of the traces. But that person needs an internet connection, so he can connect with the BVH network. It is not necessary for another person on the crime scene to have an iPad for trace administration. There is usually plenty of time to do that by the crime scene investigator. Once a while, the people need to go out to catch some fresh air and to have a break. In the meanwhile they can discuss further steps. It is also a good moment to add traces to the chain of custody. The iPad would be ideal at smaller cases like a burglary. At a homicide case, multiple persons on a crime scene is not pleasant. Especially at smaller crime scenes they can obstruct each other physically. They also have experience with annotations in 3D models. At complex crime scenes, too many annotations would be annoying when viewing the 3D image. It should be possible to disable them seperately.
The forensic department in Rotterdam use the commercial application TRIS to manage cases. But even with TRIS, the role of BVH is very important. VTSPN does not allow to directly exchange data between BVH and TRIS. The investigators would rather want it to be allowed. In the current situation, the trace envelopes go the trace administration department where it is stored in the TRIS system. Rotterdam police, and also other police units, do not take video footage that can be used in court. It turns out to be really difficult to use in court. iFEA iFEA (Fingerprint Evidence Analyzer) (image 4) is a an iPad app developed by HobbitIS that allows the crime scene investigator to quickly proccess fingerprints. Several police forces in the Netherlands participated in a pilot with this app. The participants were very satisfied with the use of a tablet. The advantage of a tablet is that it is possible to take photos of fingerprints with it. A demonstration on the iPad was not possible at that time, because iFEA did not work with the latest version of IOS. The photos can be annotated on the iPad. Because iFEA was made by the same developer as TRIS, it can interchange data the desktop version of TRIS. iFEA is slightly out of the scope of our research, but it shows the wilingness and enthousiasm of the police to use modern equipment in their work. In Appendix E, the success of the app is supported by a few reviews. The police have not decided yet if they want to implement this app in their workflow. Image 4: iFEA
Den Haag The procedure at Den Haag is as follows: The CSI first looks around on the crime scene without altering the scene. Then they come up with a plan of approach. What traces can they collect? Are there digital traces? Finally, they start to collect the traces. Just like Rotterdam, they have no standard operating procedure, but they write notes on a tablet. These data are then written in BVH. They also participated in the pilot of iFEA. The ones who used it, were satisfied. Ocassionally, panorama photos or panoscans are made, a 3D scanner is sometimes borrowed from Rotterdam. SummIT, another case management system, is also used [SummIT]. There is always a CPDU (coordinator crime scene unit), outside in a tent or bus. Roles are randomly assigned on a crime scene, but do not change on the scene. E.g., someone who takes the photos will not be assigned another role. There are always at least two persons on the scene. Sometimes specialists arive at the scene like digital experts or NFI. The investigators feel they are running back on technology. For example, they would like to have the technology to recognise license plates and faces. That is why they welcome the idea of using more technology in the future. The amount of traces collected at a case depends on the case, this can be fifteen at shootings and fifty at a stabbing case. Sometimes you will only find one trace. For example in the summer where the victim only wears shorts. In the winter, one would obviously wear more clothes and produce more traces. The CSI unit of the Hague has many kinds of cases: Homicide, vice, shootings, stabbings, threat letters, robberies etc.Burglary cases are done by assistents just like in Rotterdam. SIN numbers are not used on small cases, because its more efficient to do that at the office. BVH is not appreciated, because it is not a forensic program, but a government program. It is not user friendly. Just like in Rotterdam, TRIS is used by trace coordination team. A cooperation with TRIS, BHV and LSV (Landelijk Sporen Volgsysteem, a trace tracking system) is necessary. No GIS systems are used in Den Haag, except GPS for location determination. A helikopter can be used for aerial photos. Expertise from Paul Lucas (Rotterdam) is requested for rectifying photographs.
Feedback on PDadmin The forensic investigator was kind of enough to provide feedback on PDadmin. He suggested a to do list in the orientation tab. It should also be clear who collected the traces in the trace list and who made the agreements. Exporting data in a secured PDF with restriction options like no printing/digitally signed/timestamp would be a welcoming feature. At the tab with the case information, possible external investigators should be registered too. He suggest RFID can be used for better chain of custody, especially because tablets have RFID readers on them and SIN stickers have RFID chips in them. This could save space on CS, but also prevents the use of a laptop with barcode scanner on the CS. The advantage of apps on tablet is that you can quickly import photos taken fromtablet. Data and journals should be able to be shared securely with colleagues. Data backup should be done on police or NFI network. Sometimes traces are transferred to other people. This should also be included in PDAdmin. Furthermore, a section with personalia of victims and perpetrators is currently missing. There are some differences in workflow between police forces. Especially between major cities and smaller cities. The police is a customer of the NFI. The police will resist against an NFI product like PDadmin if it not proposed to them with care. A good link between police and NFI systems is necessary. In order to be successfull, the police should be involved in the development and decision making of the implementation of such systems. The KMAR should also be involved, since they are a potential user of such a system. Amsterdam On a crime scene there are different roles possible: There is a CPDU, which is a forensic coordinator. This is similair to other police forces. In the case of very complex crime scenes, there is always the same procedure: A big team is formed in the beginning. The advantage of this is that it is easier to send people away than to request for more personel. There is also a CS assistant. His task is to write down the agreements and keep track of the chain of custody. Usually a pen and paper are used. A computer is rarely used. Amsterdam did not participate in the iFEA pilot. It is a welcome feature in the weekends for faster fingerprint identification, because bodies need to be given back to families as soon as possible. Panorama scnanners and laserscanners are also used, but not when its raining. The advantage of 3D scanning is that it is possible to do measurements. The disadvantage is that there is lack of quality. Panorama photos are not used by justice department or partners. Because of this, the forensic investigators think it is a waste of time to take panorama photos. There is no regular photographer, the technical detectives take photos. Writing on paper is expierenced as too time consuming. They think the proces of recording the
chain of custody can be a lot faster if a laptop or a tablet is used. But the advantage of paper is that it has little chance of contamination or loss of information. A tablet can easily be damaged. The investigators are open to new technology, but only without the hassle of using wires. Just like with the previous two police forces, burglary cases are done by forensic assistents. The only difference is that they do not use the same trace envelopes. The Amsterdam team also has a lot of complaints about BVH. It is not up to date and some simple items, like a paper cup, cannot be imported. They claim that BVH is meant for police on street and not for forensic investigators. The tactical teams work with SUMMIT. On an average case, dozens of traces are collected, the alltime record is 460. Amsterdam also uses an additional system called Confisk, so they have to write things two or three times, but sometimes it is possible to import data. PDadmin feedback The following feedback points were given: ● The ‘’Searched for x” sections needs to be editable, because you can always search for something else. ● Indicative tests should be hidden until needed. ● They dont use index numbers at the police, but number signs. ● Editing of the trace container type section should be possible. NFI The NFI does not perform complete crime scene investigations like the police, but is only called in when a specific type of investigation is required, for example to collect traces from bodies or to record bloodspatters. The main task of the NFI is to collect traces and write a report. This report is then send to the public proscecutor or court. They also write down with each trace why that trace is collected (source or activity level). In the report, a selection of relevant photos are shown. The rest of the photos are delivered on a DVD. The photos in the report are sometimes annotated in Microsoft Word. The interviewee from the NFI also criticized the police for working differently than the NFI. For example, they need to write the motivation for trace collection, that is currently lacking. Just like the police, the NFI would want a system that can be linked to their datasystem. Instead of BVH, the NFI uses a system called Promis [Promis].
Literature study
Method
In this part of the study, we look at the literature for background information about crime scene information systems and 3D scanner/GIS usage in forensic science. We will also look at related systems.Results
Case management software
Outside the NFI, the CrimePad (image 5) is an interesting crime scene management system. It supports adding text and images to an organized environment. Everything can be linked together. It is also possible to work on a case with multiple contributors. Any kind of file can also be added, making the use of PDF documents and value for digital forensics possible. With the builtin log, you can see who has made changes. This way, the chain of custody can be maintained. Exporting logs and trace lists is also possible. [visionations] Image 5: CrimepadSceneCenter (SceneWorks) (image 6) Forensics is a crime scene administration system where you can merge photos and other information. It is possible to place annotations on a photo [Spheron]. Image 6: TRIS (Technische Recherche Informatie Systeem) (image 7). This application is used for trace administration. TRIS acts as an administration system, but also as a database for traces and modus operandi. These can be searched in TRIS with a faceted search option. TRIS exists for a long time, it was launched in 1995 by the company Hobbit Imaging Solutions. It was originally meant to store images of physical traces like shoe and fingermarks. Traces stored in TRIS can also be compared. TRIS is not meant to be used on the crime scene [TRIS]. Image 7: TRIS
Locard (image 8) is a forensic case management system that has been developed with the collaboration of the police. The advantages of Locard are that it is easy to share information. Locard can be used on a computer or a tablet. The data is stored in the cloud and the system can be accessed off or online. The data can be exported in Excel. [Locard] Locard has been used by the UK police for many years. It has increased productivity and raised the speed of forensic investigation. Since recently, it is offered as ‘software as a service’ to decrease costs. [Forensic mag] Image 8: Locard tablet interface Just like Locard, Blackthorn’s case management system also works in the cloud. They pay attention to user friendliness and the sharing of information. The system can be accessed through multiple types of devices [Blackthorns casenote]. An application very similair to PDadmin is iCS (iCrimeScene) (image 9). This application works on tablets and is meant to be used by crime scene investigators on the scene. The entered information can be sent directly (with a secured connection) to TRIS or printed out on stickers. iCS is suitable for both high volume crimes like burglary cases, and for complex crime scenes. iCS is currently tested in only one small police region. Sending information from iCS to TRIS is no longer allowed by VTSPN (Voorziening Tot Samenwerking Politie Nederland). [ICS]
Image 9: ICS
3D scanners
A 3D scanner is a tool to capture real life objects and transform into a digital file. In the digital file, a 3D model of the scanned object has been created. This 3D model is a point cloud [Wiki]. Point clouds are a set of combined points in 3D space that together form an object. Besides 3D scanners, there are other ways to create a 3D model. Photogrammetry is a technique where you can create a 3D model by taking multiple photos from different angles of a certain object [wiki]. A well known example of photogrammetry can be seen in the movie ‘The Matrix’, where one particular scene would not have been possible without this technique. Photogrammetry is even possible on mobile phones. A popular program is Autodesk 123Catch, which is a phone application that creates a 3D object out of many photos. [Opitz, 2012] 3D scanners used in forensic work Currently in forensic science, 3D laser scanners are used to provide an overview of large areas and small handheld (photogrammetry) scanners are used at a very limited amount of police forces in the Netherlands. This makes it possible to view the CS at a later stage without having to visit the CS again. It can be used to determine the line of sight, reconstruct ballistic trajectories, blood pattern analysis and to reconstruct crime scenes [Lukosch et al.][Buck?] Barazzetti Explains that an important additive value of 3D models are that you can reconstruct crime scenes after the crime is over and the crime scene has been cleared or altered. With a 3D model, one can try to see actions through different actors on the scene like the victim or the perpetruator. In this way, it can also be a helpful tool in scenario thinking [Barazzetti].3D laser scanners are currently used in CSI, but they have some limitations. For example, because of their static nature, some areas of a crime scene are not covered. This is especially the case indoors. The constant moving of the laser scanners is also a hassle. 3D laser scanners are mounted on a tripod and are not as flexible as a handheld scanner. Some examples of 3D scanners used in forensics are mentioned below. The Faro Freestyle 3D scanner (image 10) is a handheld device that is linked to a tablet. The user holds the scanner with one hand and a tablet with the other hand. Faro provides its users their own 3D viewer software called Scene. The most interesting feature in this program is that it is possible to clip 3D models (remove walls). [Faro] Image 10: Faro UAVs can also be used to make 3D recordings of crime scenes. They can then be used to replace laser scanners, because they can not be used inside. An example of a UAV scanner is Draganflyer. The Draganflyer (image 11) takes multiple photos during its eight minutes flight, just like a regular 3D scanner. The photos are then processed into a 3D image with the PIX4D software. Very accurate measurements can be made afterwards. Next to the the 3D model, the seperate photos can be viewed directly by clicking on a point in the 3D model. Very accurate measurements can also be made. The Pix4D software makes it possible to annotate 3D points. The 3D model can also be imported in ArcGIS. [Draganflyer Youtube][Draganfly] [Pix4D] Drones are already used by the Dutch police to take photos that can be converted into 3D files later.
Image 11: Draganfly An alternative 3D viewer with annotation option is the Leica truview. This 3D viewer uses laserscan data and it works in a browser and allows quick zooming to annotated points. It even has an embedded panorama viewer. [Leica][Example] [forensic 3D laser scan] 3rd tech (image 12) also uses 3D laserscans, but it is possible to merge multiple files that provides a great overview of the entire scene. Their software can import . They also offer the ability to save the 3D model on a DVD that can be handed over to laymen like attorneys. It is also possible to add annotations with multimedia options. [3rdtech] Image 12: 3rdTech SceneVision3D
A powerful handheld scanner is the eyesMap (image 13) from the Spanish company Ecapture. This scanner is embedded in a very powerful tablet which has the potential to create very detailed scans. It features a depth sensor and two regular cameras for stereographic photos. It is also capable of creating orthophotos. But this product is very recently developed and by the looks of the scan quality, the software is not fully optimized yet. Image 13: eCapture eyesMap A very advanced 3D viewer is Veesus Arena 4D (image 14). Besides viewing 3D point clouds, this program allows the user to not only anotate the model, but also to edit it. You can remove objects for example. Furthermore, it allows various view options like first person mode. Image 14: Veesus Arena4d
Forensic use in the Netherlands The visualization team from the Rotterdam police is one of the few units in the country that is in possession of a handheld 3D scanner and they have used it to scan a baby corpse. It is already possible to place hyperlinks with detailed images of that object in annotations in a panorama photos. An option like this for 3D scanners would be very welcome, since the resolution of a handheld 3D scanner is too low to observe details.
3D Scanner Technique Accuracy Cost
DotProduct Photogrammetry 1 mm (close range) and 15 mm (long range)
$5150
Faro Laser scanner 1.5 mm Unknown
Draganflyer Photogrammetry 2 cm (99%) Unknown
123Catch Photogrammetry Reasonable Free
Leica truview Laser scanner Very accurate Unknown
eCapture Eyesmap Photogrammetry 10mm@1m €8000 (estimate)
Comparison of some 3D scanners Advantages 3D scanners can be used in crime scene investigation to quickly capture the whole crime scene. Compared to the traditional method of taking photos, a scan can be done within a few minutes. The advantages of a 3D model is that you can place annotations at an exact point in the model (x,y,z coordinate). In this way, the viewer can always see the annotiation, regardless of the point of view. This technique is called billboarding. A 2d image, in this case the annotation, always faces the camera, just like a billboard faces the drivers on a highway. This technique is usually used in 3D applications like games to gain performance improvements [Wiki Sprite]. 3D models can help the people in court understand to interpret the crime scene. 3D objects can be viewed from different angles [Komar et al.].
Limitations 3D scanning is not a perfect recording method. The quality of the scan depends on the device used, but also on the skills of the user. Without proper training and experience, artifacts can occur in the 3D image. This is especially the case with handheld scanners, where the scan quality depends more on the users movements [Komar]. Due to memory limitations in DotProduct, it not possible to scan very large areas at once. But it is possible to stich different 3D scans in a program like ReCap [stiching videos]. In most 3D viewers, it is possible to measure the distance between two selected points (x y z coordinates), but it is almost never as accurate as real life. Therefore, it has limited value in forensic cases.
GIS
Research from van Crugten suggests that merging photo’s together with the ArcGIS program can provide an overview of the CS. With the programs ArcReader and HTML Image Maps, these maps could be viewed best. A Total Station is required for the process. Further research needs to be done to measure indoor performance, since GIS is originally meant for outside 2D measurements. A GIS (geographical information system) is a system which makes it possible to present geographical data. The aim is to provide the viewer specific kind of information. This information can be presented as layers on a 2D or 3D map. It is also possible to combine different information types and present them at the same time. Everyone who went to school will remember the geography books. GIS is no different than those, except the content is digital and can be modified this time. GIS is used in many sectors. The most popular applications of GIS are maps like Google Maps and GPS navigation systems. It is also used by professionals in environmental planning and by military commanders. [GIS wiki] GIS can also play an important role in finding hotspots in an area by combining several layers of information on a map. Such an approach has been used in a study of metal contamination in an urban environment. [Li][Gong} GIS is increasingly used in forensic science. It can be used as a helping tool in forensic geography (also called forensic mapping) or with the creation of an overview map of a crime scene by combining several overlapping images. [Alfhadli] In the case of forensic geography, GIS can be used to provide an overview of outside crime scenes or as a tool to present location data. Location data can be obtained from GPS or cell phone activity. It is used in court as evidence in some South African high impact cases. [Schmidtz]
Data from mobile phones can provide information about the presence of a certain suspect at a scene and can tell something about their activities. These activities may involve visiting certain places like the house of the victim or a remote area where a body has been buried or found. By using cell phone data, alibis of suspect can be checked. The interesting thing about GIS is that they are easily understandable (by judges) and they present evidence in a nonbiased way. [Schmidtz] It is also used in the Netherlands in the Deventer murder case, where it provided an overview of the larger scene regarding cell tower data. There is plenty of GIS software available, but ArcGIS remains popular at the NFI, altough they are looking for an affordable alternative. The NFI also wants to use GIS to develop an application called SCAR (Systeem voor Calamiteiten Registratie) (image 15) in cooperation with the police. The aim of the application is to provide an overview of traces in the case of major disasters. This application makes it possible for forensic scientists on the scene to send photos of traces anotated with a SIN and exact location to the command vehicle. In the command vehicle, a database will be present and a GIS will be used to create a map of the scene. The project will start in september and it is scheduled to be finished in early 2017. Image 15: Concept of SCAR Recommendations In summary, GIS is more suitable for overview purposes with outdoor crime scenes, altough it can also provide an overview at indoor crime scenes, for example with the purpose of scenario thinking. By placing overlapping photos next to each other, a seamless raster photo can be
formed. This technique is called mosaic photography. It is similair to aerial orthophotography. The advantage of this technique is that you have a top view of the crime scene and can support hypotheses. For example, what path did the victim or perpetrator take?
User tests
Method
Aim The next step in our research was a test of some currently used or experimental registration techniques in an extensive user study. The aim of these tests is to research what the value is of these new methods are in terms of registering a crime scene and the experience of the enduser. As mentioned in the introduction, we have chosen PDadmin and DotProduct as an example of a case administration tool and respectively a 3D scanner. We will compare these techniques to the classic registration and panorama photos. This test will simulate the ability of a judge or someone from the justice department ( e.g. lawyer, judge) to get an overview of the crime scene. The results from different crime scene registration techniques will be compared. For these tests, participants will be divided into two groups. In every group, they will evaluate two techniques.The second test will always take place in another room of the fieldlab to prevent prior knowledge and bias. For an extensive description of the techniques, see below. Explanation of crime scene registration techniquesClassic (paper form + anotated photos (on laptop)) (Appendix E): Traces are written on an NFI form, see The SIN stickers are attached on this form. There is also a collection of crime scene photos. These can be divided into two groups: overview photos and trace photos. In the traditional method, a close up photo of the trace, a medium shot and an overview shot from a distance is taken.
PDadmin (application on laptop + photos) (Appendix F): Photos that are taken according to the current forensic method used by investigators. The photos will not be anotated, since the SIN numbers are already visible
Panorama (interactive viewer on laptop + anotated photos) (Appendix G): Is used to provide an overview of the scene. It is the first thing that is showed in court or to the crime scene unit. The advantages are that you have a very good overview of the scene and you can see how the locations and objects are connected. You can also look up to see the ceiling. The panorama can be viewed in a browser. It is possible to switch to another view by clicking on an icon on the map. The disadvantage of a panorama photo is the lack of meauring distances
which is possible with laser scans. The DotProduct scan is usually not very accurate in measuring distances. It is not possible to annotate in the panorama viewer, but it is possible to do that with an external photo editing program. The annotated image can then be seen in the panorama program. It is also possible to include a link to a detailed photo, but this is never done, so this will also not be done in the test setup. Police units that have a panorama camera, use it at every major crime scene.
3D viewer (on laptop) (Appendix H): An anotated 3D scan of the crime scene viewed in Autodesk Recap. The 3D scan has been created with a DotProduct 3D scanner. Sample size The test will be a comparison between four techniques. In accordance with a statistics expert at NFI a sample size of fourty has been chosen as sufficient, on the condition that major differences are expected between the groups. We expect major differences between the different registration techniques. We will then apply a Ttest to find out if the accuracy rate of two different methods are different. Image 16: Sample size requirement Preparation A fictitious scenario with multiple types and amounts of traces was created.
With the help of experienced CSI, the house was ‘decorated’ with numerous blood stains and spatters. On top of that, additional traces like 9mm cases were placed on the crime scene to match with the scenario (image 17 and 18). Target Traces Test 1
Image 17: 21: Cigarette bud , 23: Suitcase , 24: Shoemark in blood , 26: Safe , 34: Ducktape
Image 18: 1: Pizza box , 8: Bloody knife, 9: 9mm case , 13: Bloody closet handle, 18: Memory card case Test groups Test 1: Two groups: Classic way and PDadmin (bedroom) One team will try to track the traces in the traditional way and the other team will use PDadmin. Group A: Classic method (20 participants) This group will be given a laptop where they have access to a folder with all the crime scene photos taken by the crime scene investigator. The group will also receive the paper trace list as it is in the current way. Group B: PDadmin method (20 participants) This group will be given a laptop where they can use PDadmin. The content of the information will be the same as group A. They will also have access to the photo folder. Test 2: Two groups: (Panorama and photos) + 3D (living room/kitchen) One group will only be presented photos and one group will see the 3D scan. Group A: Panorama + photos + paper (20 participants)
These participants will be the same as group B from test 1 (PDadmin). This group will be given a laptop where they have access to the panorama photos with a map. They can also see the crime scene photos. The group will also receive the paper trace list. Group B: 3D (20 participants) These participants will be the same as group A from test 1 (classic). This group will be given a laptop where they can use Autodesk ReCap. Here they can see a 3D model of the scene with anotations of traces. Scenario The house is occupied by a drug dealer. At one point, someone from the competition dressed up as a pizza delivery guy showed up at the door with a pizzabox in his hands. His aim was to rob him of his drugs and money. When the house owner opened the door, he rushed inside and had a vicious fight and in that fight a few shots were fired and the house owner was stabbed. This resulted in a very bloody crime scene. Participants The participants of the tests were mostly personel from the NFI working at various departments. This was not an issue, since these techniques should be understandable by everyone. It was a mixed group, so there were people from different ages and genders. Registration The mock crime scene was registered in different ways by two crime scene investigators from the Rotterdam police department. A paper trace list was filled in by them. It was converted in a digital trace list into PDadmin by me. The regular photography, panorama photography and the 3D scanning with a handheld 3D scanner was also done by them. Instructions The participants have the opportunity to read the written instructions (Appendix b) prior the tests. Additional instructions are given prior to each different test method where the participants could ask additional questions about the method. The participans were givens some time to get used to the panorama and 3D viewer controls. When they tought they were ready, the tests started.
During the test The participants were also asked to think out loudly, so their real time experiences could be observed. Locating the traces by the participants was done by placing markers with the numbers of the particular traces. The use of other ways of locating (stickers or a marker) were not preferred to avoid contamination of the trace locations. The participants had a maximum of five minutes to finish one test. A questionnaire to evaluate the used method was filled by the participants afterwards. Thom Schroder has performed a user test where two groups had to collect traces with the above methods. The difference in my tests are that the groups will not collect traces, but they have to locate their original locations. Traces For every test, five trace locations were selected. We tried to select representative traces that differ in difficulty. So some trace locations should be easy to point out, but some are very hard. The crime scene has been split up to two parts for the experiments. Test 1 will be in the bedroom with five traces out of twelve. Test 2 will cover the entrance, hallway and living room/kitchen. We selected five traces out of twenty.
Measurements The following things will be measured: ● Time (Time will be measured in seconds, there will be a limit of 5 minutes) ● Accuracy (is it within a reasonable range of the original trace location?) ● System usability by a questionnaire [Brooke] ● Observations of participant behaviour by the researcher The participants that tested the techniques will be asked to fill in a questionaire about their experience. In this questionaire (Appendix C), they will be asked to rate the system by assigning a score at different statements. They are able to provide a score of 1 to 5 (Likert scale). 1 meaning ‘I completely disagree’ and 5 meaning ‘I completely agree’.
Method Classic PDadmin Panorama 3D
Trace list Paper Digital list Paper Annotated
photos photos photos embedded Overview of the methods
Results
Time 40 respondents participated in the experiments. The results can be seen below. Since some users tried to perform as fast as possible and some users utilized most of the five minutes they had, we will not use the time (image 19) as a factor to compare the methods. A pattern of correlation can be recognized in the below two graphs. We calculated the Pearson correlation to be sure and found a correlation coefficient of 0.9, which indicates a very strong correlation.Image 19: Average time of experiment completion Accuracy
Image 20: Average accuracy of marking trace locations We have performed a twotailed Ttest to find out if the accuracy (image 20) differs significantly between two methods. We first look at the difference between the classic method and the PDadmin method. We use a confidence level of 95% which results in an alpha value of 0.05. [Sauro] Hypothesis 0: The accuracy of the two methods differ significantly. Hypothesis 1: The accuracy of the two methods do not differ significantly. With the statistical program SPSS we have calculated a value of 0.061. 0.061>0.05 We reject hypothesis 0, the result is not significant at p < 0.05. This means that purely on this experiment variable (accuracy), we can not claim PDadmin is a better registration method. With a larger amount of participants, the result could have been significant. Secondly, we look at the difference between the panorama method and the 3D method. Again, we use a confidence level of 95% which results in an alpha value of 0.05. Hypothesis 0: The accuracy of the two methods differ significantly. Hypothesis 1: The accuracy of the two methods do not differ significantly. With the statistical program SPSS we have calculated a value of 0.316. 0.316>0.05
We reject hypothesis 0, the result is not significant at p < 0.05 System usability score According to the system usability scale by Brooke, different weight scores were assigned to the statements in the evaluation form. The weight of a positive statement has the following calculation: scale 1. The weight of a negative statement will be: 5scale. The sum of the scores will be multiplied by 2.5. Statement Weight
I found this method pleasant to use score-1 I found this method easy to use. score-1 I quickly realised how to use this method. score-1 I think that this method can be learned
quickly.
score-1 I think most people can easily use this method score-1 I think the chances of errors are very high with
this method.
5-score I was confident when I used this method. score-1 I think I performed quickly. score-1 I think I performed accurately. score-1 I think I performed well. score-1
Image 21: The average score according to the user ratings System usability score During the observations, I noticed that the PDadmin group never looked at the original photos because they did not seem to need them. In the panorama group, most of the people only used the panorama viewer, but a few people skipped it and viewed the trace photos. In the 3D group, some people had major problems with navigating and ended up in a perspective without overview. They then tried to localize the traces with the annotated photos only. Generally, people who were older or had little experience with technology had more difficulty to get used to the 3D controls. We also looked per method to see in detail how the participants rate the methods. We can see that PDadmin generally scores higher than other methods. The participants especially like how quick they can learn to use PDadmin. Furthermore, they think that the classic method has the highest chance of errors, which is also confirmed by the lowest accuracy rate amongst the methods. Classic
Image 22: Classic ratings PDadmin Image 23: PDadmin ratings Panorama
Image 24: Panorama ratings 3D Image 25: 3D ratings
Classic PDadmin Panorama 3D scan Positive points Easy and simple. Paper and photos are familiair for everyone Photos linked to traces, everything in one program, quick Good overview of the crime scene. A lot of freedom of movement and zooming Very accurate overview and locations of all traces Negative points Bad overview, too many photos to search through. Time consuming Photo album and interface Takes a long time to get used it. Handwritten information on paper. Difficult to use. Quality of 3D model. Improvemen t suggestions Photos in subfolders per
trace. Better descriptions Interface, more linked photos, link with map/3D Modified controls. Embedded tracelist. Traces annotated in panorama. Better resolution. Faster computer. Userfriendly controls A summary of the feedback on the various methods by the participants We also registered if the participants visited the CSI lab before. The reason for this is that there could be a possibility that they would have an advantage compared to the first time visitors. But we did not notice differences in the time spent or in accuracy.