False/missing features in live scan rolls

[bficken]

Hello Everyone,

Does anyone know of research articles on false features created by the process of rolling a print via live scan? I'm looking for studies that demonstrate frequency with which non-existent minutia are added or existing minutia are dropped in rolled live scan prints. I've got one from 1988 but that was a *little* while ago ha. Thank you!

[Boyd Baumgartner]

It's called an image mosaic, when slices of the rolled image are stitched together via algorithm.

Here's an entry article. I'm sure you can find more based on papers this references or ones that reference this paper.

http://citeseerx.ist.psu.edu/viewdoc/do ... 1&type=pdf

[josher89]

Is this similar to what is happening when you register a finger on your cell phone? Just splicing the pieces together?

[Shane Turnidge]

These optical artifacts are unfortunately a big problem for LPEs.
Here's a link to my website for some insight.
https://sstforensics.com/live-scan-issu ... artifacts/
Shane Turnidge

[Mike French]

I've seen quite a few of these mosaic artifacts over the years. However, my observation has been that the vast majority of these images come from outdated livescan equipment (more than 10 years old). I'm pretty sure that Moore's law has allowed reputable manufactures to virtually eliminate the issue. Is anybody seeing mosaic artifacts from newer equipment?

[Shane Turnidge]

These artifacts were from relatively new machines. (Circa 2013-2014) or so.
At the time we believed that the person being printed was ever so slightly moving in a way the machines couldn't handle. As you can see, the images look fine until you drill down on the analysis.
It was a latent ID that brought these images to light. FWIW I don't think this issue is particularly common but I do wonder how many images like these are out there yet to be discovered or will their discovery be buried.
Shane Turnidge

[Mike French]

I like the fact that the RCMP includes visual examination of livescan outputs as part of their certification process. FBI appendix F certification only tests the sensor capability, but without visual examination of the images, these types stitching artifacts would never be found in government certification. I think its safe to assume modern LS technology is able to produce high quality images with very low frequency of these types of artifacts. But software development is only as good as the QA program of the manufacturer which I am sure varies from manufacturer to manufacturer, and even varies from the same vendor over time as priorities and budgets change.

[Shane Turnidge]

Mike, I think there's a lot to be said for human QA processes on the front end of a live scan acquisition. When print takers are given superficial training and the ability to over-ride the transaction this type of issue can arise in spite of the software and hardware. Technology does its part in the process but people are still integral to a quality known print database. Managers should encourage their print takers and give them the tools they need to do a quality job. Print taking is not an easy job but it can be supported with good managers who understand the issues.
Shane Turnidge