Public R&D-Projects

Current

European Project : FIDIS - Forensic Implications of identification systems.

Past projects

 

Use of biometrics in forensic science

 

Validation and research to image processing methods on finger marks.
This project should answer the questions : which image processing methods are acceptable and what are the limits of old methods (FFT, dilation and erosion, subtraction) and the new methods, for example with wavelets. In several court systems image processing methods are accepted, however the limits are not defined.

Image Databases

The comparison of commercial image databases and their way of correlation.

Correlation in a database of cartridge cases

This research is a comparison of correlation methods. Furthermore the different kinds of wavelets (Gabor, a trous etc.) are compared for preprocessing the data in the database.

Database of logo's on Drugs pills (use Netscape to enter this database)

This research is a Senter project (GIDEON) conducted together with the University of Amsterdam. We have compared our correlation techniques with the software used in Amsterdam (Zomax). Furthermore the logo's have been acquired with structured light of OMECA in Berlin. Please find a Master Thesis of Jitteke Struik on the use of this system here.

Correlation of toolmarks
This research will compare the methods that have been used in our previous publication on 3D-image data which is acquired with structured light. Furthermore in a joint research project with the University of Gliwice in Poland, we have investigated the use of neural nets in these kinds of databases. Information about the results is available here.

Super Resolution
The different approaches for super resolution need quality control. We tested a sequence of JPEG-compressed images, and it appeared that the wrong conclusions could be drawn. This research is conducted together with the University of Queens in Belfast, Northern Ireland.

Identification of camera's
In child pornography cases it can be important to find a relation between camera and image on the internet. There are several ways to do this. With several cameras we could find pixel defects. Please find more information here. This method has worked succesfully in several cases.

Gait Analysis in Video

To investigate the possibility of time-lapse video, time average hip joint angle (am) is also analyzed in time-lapse video (12.5 Hz). A small decrease in the number of significant differences between subjects was found (from 58 % to 56 %). This means that for a time-lapse video is useable for recognition purposes.

Bulletcytology

This project is conducted together with the pathologists from the Netherlands' Laboratory of Forensic Pathology. In this project we are doing some research how tissue will transfer to bullets. In many cases it is possible to determine which part of the body a bullet went through by the tissue which is left on the bullet.

 

CT-scan for caliber determination

This project has been carried out together VU-University in Amsterdam. We have  tested with raw data from a CT-scan if it is possible to determine the caliber of a bullet in a human body. It appears to work with a special setting of the CT-scanner, to avoid streak artefacts.

 

REBEZO : storage of shoeprints

REBEZO is a project for the storage of shoeprints in a computer. In practice shoeprints are a very valuable piece of evidence at the crime scene. It is often said that this is the most overlooked evidence at the scene of crime. With this database it is possible to store images of shoeprints found at the scene of crime, to collect information of shoeprints of suspects and a reference database of shoe profiles.

 

TRAX : storage of toolmarks

TRAX is a project for the storage of toolmarks in a database. Often with burglaries some striation marks or impression marks from screwdrivers, crow bars or pliers are found. The images of this information can be stored in a database of images. We also developed some methods for correlation of striation marks in such a way that it will be easier to find the right match in the database.

 

Drugfire / IBIS

At our department we have a high caseload of spent cartridges and bullets (and also a huge backlog). We have evaluated both systems. Currently we are switching to IBIS since Drugfire is not supported anymrore.

 

3D-scanning

It is expected that in future the microscope will be partly replaced with some 3D-scanning equipment. On the market there are different methods for laser scanning. We tested some of those systems, like the laser profilometry system. We have tested the structured light methods from OMECA for obtaining a fast 3D-image. The problem with laser profilometry at this moment is that it is slow and that there are some artifacts in the scanning. A laser scanning system (with polygon mirror etc) might solve this problem.

 

Drugs-pills database

When developing the REBEZO and TRAX-system there was some interest of the Drugs department of our laboratory for storing XTC-pills in a database. This development has been a combination of Delphi and Borland C.

 

Image restoration

Image restoration is often asked for video's which are not very sharp. In this research project you can find more information on deblurring techniques, as MTV, Maximum Entropy and Wiener Restoration.  Often however the quality of those videos is not very good. Restoration will not always result in good results. For this reason the best solution is to have a good recording system combined combined with a good camera system and light. Digital systems might also work convenient, and the challenge is to have an authenticated system.

 

Photogrammetry : Monte Carlo and animation

This is a field that seems to be easy, however in practice it is easy to make mistakes with real world video systems.