Sunday, July 20, 2008

July (4)

This month is really interesting in forensic way. Last Thursday I had to testify in court on a face comparison case. First I had to wait for four hours outside of the court room, before I was called to testify (these are small disadvantages of being a forensic scientist, an expert often has to wait for long times in court, however Dutch courts tend not to call expert witnesses as often as in the United States).

It appeared that they were ready for asking the questions, since all three judges, the two lawyers and the prosecutor were asking questions. It was slightly more difficult compared to other cases, since I was not allowed to have my notes and the reports that we had written in front of me (they took them away). I just had to answer the questions, without looking to any written material.

Then the questioning started. The usual questions were asked concerning expertise, however now there were more questions. We had written our report in a Bayesian conclusion, so I had to explain how this worked, and how it developed. Also some nice questions were concerning the software we used (image visualisation software) and if it was validated, and how it was validated. Furthermore questions concerning proficiency tests between different laboratories and how the field works.

Of course it would be nice if face comparison would be more objective, then with the method that we use now with three different examiners filling in sheets concerning the comparison of the different parts of the face. However the issue is that there is not enough solid research to make it more objective.

This is in many fields in forensic science the case, that the experience of the expert counts for the conclusion. More research should indeed be done in these fields, to make them more objective, and it was for me somewhat disappointing that the European Commission did not put this in the seventh Framework Research projects yet, however I understand they have to make a selection. In the past several efforts have been made in European projects, such as ear comparison fearid etc, however it appears that we need more solid research on the different fields in forensic science to make the conclusions more objective and make the different conclusions of different experts become more calibrated.

Saturday, July 12, 2008

July (3)

Some software is more complicated to develop then other software, and sometimes it takes much more time then expected, however the results count. With one of the projects were I am project leader of, defraser, it appears that we have now a stable product which can be used to analyse MPEG-1, MPEG-2, MPEG-4 and 3GP. Broken video streams can be analyses as well as many others, and since it is open source others can also develop on it. Students from the Hogeschool Amsterdam have developed plugins for AVI and JPEG, and we expect them also to be available soon. You can download it for free from https://sourceforge.net/projects/defraser/ and of course we are more than happy if you can also develop plugins. The software works for broken or parts of 3GP-files for example from mobile phones.

Monday, July 07, 2008

July 2008 (2)

Just for two days in Brussels at a review meeting for the FIDIS www.fidis.net project at the European Commission. It is nice to see which deliverables have been made in this project. Also got some time to do so planning, en even to update sourceforge with our newest PRNU tool for identifying cameras in Java. It can be found at https://sourceforge.net/projects/prnucompare/ and is made by a student Maarten van der Mark from the Haagse Hogeschool and I was supervisor for this project. It is nice to see how we can develop nice tools in Java, however validation remains important.

One should take many flatfield images (white/grey area) to determine the PRNU. With this possible a small database can be made, and several cameras can be entered. We also would like to add this software to a deliverable of FIDIS, where we try to link cameras on Youtube or other sources.