|
|
 |
|
Q&A
|
 |
|
 |
 |
Forensic video analysis
May 6, 2008
|
|
by GV staff
|
Q: How complicated is forensic image analysis? How long does it take to become proficient in or master this skill set?
A: Forensic analysis of any type is incredibly complicated. Yet, every day there are well meaning folks in law enforcement receiving budget/grant money and buying the equipment necessary to perform the tasks. Some have no experience whatsoever. Some come from backgrounds that strengthen their beginning position.
From a video/image standpoint, politicians across the country have been cutting ribbons and placing all sorts of CCTV systems in their cities. They often have little understanding of the back-end support necessary to turn the footage that these new systems generate into actual viable exhibits suitable for court.
Consequently, agencies often do the best that they can with what they have. As far as mastery is concerned, I've been in the art world for almost 30 years and I'm still learning new things. Just when you think you've got it, a new device or a new piece of software comes out and changes everything. Mastery is more of a journey than a destination.
Q: How important are standard operating procedures or a documented workflow for forensic image/video analysis?
A: Documenting your workflow is likely more important than a set of SOPs. To meet the court's standard for this type of evidence in many states, we must produce reliably repeatable results. The workflow helps in this aspect. Machines and software may change, I may emphasize one aspect more than another, but the overall workflow remains constant. I have SOPs, but they tend to be generic so as to accommodate for the wild shifts that technology takes.
 |
| Jim Hoerricks, Senior Forensic Video Analyst, Los Angeles Police Department Scientific Investigation Division; Author, Forensic Photoshop |
Q: Critics argue that forensic video analysis changes rather than clarifies an image. How can you describe the process in a way that critics or juries will understand?
A: The critics are right. I change images all the time. The key is looking at what changes and how it is changed. The main difference has to do with content and context.
When you woke up this morning, you probably turned on the light in your room. Suddenly, things that weren't visible in the darkness sprang to your awareness. Did the act of turning on the light place the chair and dressing table in your room? Hardly. The act of turning on the light illuminated the room and allowed you to better see what was there.
That is what we are doing with images. The contents of the room remain the same -- the chair, the table, the bed. The context remains the same -- it's still your room. Add to all this the fact that we work on copies of the original media -- preserving and protecting the original for cross examination.
Q: Which is better source for forensic video analysis, analog or digital video footage?
A: Better is relative. If faced with a choice of a VHS tape from a CCTV system and digital files stored at 1CIF and 1 fps, I'd take the tape any day of the week. For me, I'll say that digital is better when it's parameters pass those of NTSC video. Less than 525 lines / 30 fps and you may have missed something. There is some amazing stuff out there that utilizes multi-megapixel cameras and uncompressed storage. I'd take that over VHS any day.
Q: Are there any codecs or compression technologies that should be avoided for surveillance video? Why?
A: The discussion of codecs and compression always causes the most controversy. Once digital CCTV evidence hits the courtroom, the expert with the best forensic skills often wins. In this context, I mean forensics as debate and oration. There are some excellent speakers out there who have been able to convince juries about problems with digital video evidence. Some attorneys never even bring up potential problems, perhaps through a lack of knowledge of the issues.
For the most part, the objects are the following:
Interlacing. Some systems/formats still employ interlacing. When both even and odd lines are preserved, you will find less objections. When only one set is preserved, and the computer interpolates the other, then the objections start. It leads to problems with identification when half of the face or license plate may be generated from a sample of the surrounding pixels. Also, tearing can confuse a jury when heads split from bodies and so forth.
Keyframes vs. predicted frames. A keyframe can be considered a true and exact representation of the observed scene in most cases. Other than compression factors, there's little difference between it and a digital still photo. Predicted frames are much different. The computer predicts what may happen, then corrects itself. Can these be considered true and exact? Some have argued successfully against their inclusion as such.
Transcoding. This processes takes one format and recompresses it into another format, often for ease of viewing. In one case that I can recall, it was argued successfully that the attorney's DVD was not appropriate for viewing by the jury as it was recompressed (changed) from its original proprietary format. The attorney had the original files, but lacked the ability to play them back in court.
For our purposes, whichever codec compresses the least, preserves all frames, and allows for playback on most systems will be considered the “best."
Q: What type of analysis are you asked most often to do? What’s the most unusual analysis you ever had to attempt?
A: Most of our work centers around identifications, individuals and objects. These revolve around "get me a good face for a flyer” or "get me a license plate that I can run." I've worked on many cases comparing tattoos from surveillance video to booking photos, as well as getting a face from a reflection in a nearby window.
Q: How can you reduce the “CSI Effect” with juries?
A: The best way to combat the CSI effect is for the attorneys and experts to work closely and plan the presentation of the exhibits. It's important for the attorneys to use the experts to "frame" the discussion of the exhibits. This again is forensics as oration/discussion. I have worked for the studios, so my opening would work something like this:
"In the movies and on TV, the script is written and the filming begins. The actors, the director, and the editors all know how the show ends. They all know who did it. They can use special effects to dramatize the revealing of details already known to them -- adding music to heighten the drama.
“In police work, we don't have the benefit of working from a script. We don't know how the movie ends. We take each piece of evidence as we find it, putting the case together with what is presented to us. The facts of the case lead us to a conclusion, not a predetermined script. With that in mind ... "
In this way, I've taken the technology completely out of the discussion and reframed the debate. We are now in the world of what we know and what we don't know, rather than Hollywood's make believe.
|
|
|
Government Video is part of the NewBay Media Community
copyright 2007© NewBay Media LLC.
810 Seventh Avenue, 27th Floor New York, NY 10019 T (212)378-0400 F (212)378-2160
|
|
|
