Flatbed Scanners vs. Digital Cameras...

[SConner]

Anybody out there aware of or know of any studies that have been done comparing the quality of flatbed scanners vs. digital SLR cameras when it comes to capturing latent prints for comparison purposes? Anybody out there who wants to express an opinion or share an experience using the two also please feel free. My agency is looking to incorporate flatbed scanners as an option for capturing friction ridge detail and I'm looking for reference material if it exists.

Any help provided is greatly appreciated in advance.

[George Reis]

I don't know of any studies - especially relating to comparisons of latent prints captured both ways. Also, the resolution of even lower-end digital cameras and flatbed scanners exceed what you need to capture third level detail in latent prints.

That said, I have always preferred capturing latent print cards, ninhydrin prints, etc. with a flatbed scanner. The light is consistent, the capture is done with three separate samples for red, green and blue, and the images are already sized to 1:1 when scanned at 100 percent. In addition, scanners can frequently pick up detail in 3-D objects that is otherwise difficult to photograph because of the lighting angle of the scanner. I've seen some excellent results with footwear and with latent prints on shiny surfaces, including handcuffs, from scanners.

Of course, you won't be able to capture fluorescing prints using the flatbed. It would be interesting to modify a scanner to include multiple light sources and linear arrays that include cutoff filters to provide multiple outputs with a quick scan - this is something that would be beneficial for latent print examiners and questioned document examiners alike. Imagine a scanner with white light, and perhaps five or six other wavelengths, and a CCD that had six or eight rows of 1-D pixel arrays, one with no filter, and the others with various cut-off filters. Perhaps this could include UV and IR. Then each scan would have the ability to create a multi-channel output giving the results from each linear array. Do six scans, and you have 36 results (assuming six light wavelengths and six rows of pixels). The driver would automate the process - allowing a one button click to get any or all of the combinations that one set.

Sorry for my digression. This is something that I've thought about for some time - and get a bit excited about.

Back to comparing scanners to cameras - both will do an excellent job, but the convenience of the scanner and the consistency of light quality make it my preferred choice when appropriate. It's an additional tool that is frequently used in my lab.

George
Imaging Forensics

[Neville]

I would look at it a bit differently, I think you have more control over a camera and the lighting of the object, giving more depth of field as you need it with the f stops, most work can be pushed through quicker as most things require little change to the settings, there is the option of back lighting and using filters with the top lighting. Most flat beds I have used are just a bit slow for me although I have never used high end flatbeds. The cost of SLRs are coming down but you must get high quality lens though, what you can save on the MPixels spend on the lens.
I know ninhydrin prints come up very well on flat beds but most work is lifts from Scenes of crime here in NZ, the ideal would be surely to use both.
Thats my opinion for what it is worth.

[Shane Turnidge]

There's no doubt a camera gives you much more flexibility when it comes to grasping a difficult image. You also have the additional advantage of enhancing the image in any number of applications designed for the purpose. But before you give up on the flatbed scanner, particularly if it came with your AFIS system, consider trying this.
For those latent prints that are really faint, try turning the latent 45 - 90 degrees from the original scan. Sometimes by scanning the impression from a different vector you can obtain better contrast, perhaps enough to encode even the toughest "ghost" print.

Shane Turnidge

[mdavis]

Actually, the depth of field of a flatbed scanner is much deeper than needed for a simple lift card. This is attributable to both the effective lens opening used and the brightness of the light wand used. Some artists use scanners to capture bas-relief objects (flowers for example) for their work. Many flatbed scanners also use the older CCD sensors which are largely abandoned in modern digital SLRs due to their higher power requirements and subsequent battery drain considerations, not an issue for an AC powered scanner. CCD sensors are considered to be perhaps very slightly "sharper" than CMOS sensors which is why Nikon stubbornly resisted changing over until Canon waxed them on battery life, causing competitive pressures that forced the change. Nonetheless, many flatbed scanners scan at a "native" resolution of at least 1200ppi and many now exceed that several times over. A typical APC sensor (1.5 or 1.6 crop factor) in a digital SLR rated at 6mp capturing a single fingerprint with scale and ID information will exceed 1200ppi resolution by about 10-20% depending on the image size.

For my work, if fine detail is needed and the surface is relatively flat, using a scanner is usually easier than setting up a camera and messing with the lighting. On the other hand, if off-axis lighting is a requirement (oblique, axial, etc.) then a camera is the only option. Both scanner captures (.TIF images) and camera captures (.RAW or .TIF) can be enhanced in suitable digital imaging software in the same manner. The real issue is how the image needs to be lighted to be captured with maximum detail and contrast. Since scanner lighting is even across the image, it is my preferred method when it can be used because it allows more effective enhancement than fighting with light reciprocity issues in off-axis camera lighting.