UPDATES ON CLPEX.com

No FIG update this week due to technical difficulties in re-hosting CLPEX.com to new servers.  I hope to have this issue resolved next week, but it may be longer.  On the bright side, the forum has been migrated and is operating normally!

Inserted KEPT #10- the Analysis Phase of ACE-V:  Discuss this topic on CLPEX.com - a discussion has been created for KEPT.

I have accepted a new position in the Concepts and Technology branch of the DoD Biometrics Task Force.  Although I have enjoyed standing up and managing operations divisions for DoD Examination Services, these units are operating efficiently and I feel it is a great time to move toward the tip of the spear into research and development to support the methods and capabilities of tomorrow.  There are exciting projects in the works and I consider it a privilege to transition into the management and support DoD strategic initiatives and latent print related projects.
_________________________________________

Cully Stimson brought us an excellent article regarding the success of the fingerprint discipline.

we look at a CLPEX.com forum discussion on the best resolution to use for printing latent fingerprints.

We also look at new R&D opportunities within the Department of Defense.
_________________________________________
Finding the Best Input Resolution for Printers
Author Message
Steve Everist
Site Admin

Posted: 03 Mar 2008 02:47 pm Post subject: Finding the Best Input Resolution for Printers

I stumbled across the website below while searching for information concerning the best resolution input (ppi) relative to a printer's rated output (dpi).

This has become an issue as many offices have switched over to digital capture of their latent prints, and evidence in general. A common question seems to be, "What is the best printer for printing out 1000 ppi latent prints at 1:1?" With the current crop of printers, this just isn't going to happen. Although printers advertise outputs in excess of 1440/2880/etc... dpi, they aren't able to resolve, with the eight or nine inks available to them, the color and clarity of an image containing as many as 16.7 million colors with a resolution of 1000 ppi or greater.

So the issue to consider is what resolution at input creates the the most accurate output from your printer. I've heard different ways of figuring this, including elaborate math problems with length and width dpi of the printer and the square root. I think the solution is more simple than this after using these test images with our Epson 2200.

I printed both of the images (one vertical lines and one horizontal) on our Epson, using the same printer quality settings that I use for printing out high resolution images. And it turns out that 180 ppi and 360 ppi seem to be the best. I would guess that 720 would also be good, but the printer doesn't appear to be able to resolve this level of detail as the lines are too fine to see that they're lines. The Epson 2200 lists an output of 1440x2880 dpi and both 180 and 360 are divisible into both numbers (as is 720). The clarity of the lines at 180 ppi are actually better than those at 200, 240, 260, etc... until you reach 360.

So if you'd like to test your printer to see which input resolutions seem to hit the sweet spot, go to http://www.inkjetart.com/tips/ppi/ and download the two PDF files. Remember when printing to not use the "fit to margins" setting, otherwise it will modify the image for printing. Also use the settings and paper that you use for high quality prints.

I think the diagonal lines are very telling. With the Epson 2200 at 480 ppi input, there's a block pattern created that was very interesting to say the least!
_________________
~Steve E.

--------------------------------------------------------------------------------
Gerald Clough
Posted: 03 Mar 2008 03:34 pm Post subject:

I found one guy who did extensive testing of the 2200 and found 288 ppi to be optimal. http://www.rags-int-inc.com/PhotoTechStuff/Epson2200/

Good discussion, too, and a good starting place for those setting out to analyze their own hardware.

--------------------------------------------------------------------------------
Steve Everist
Site Admin
Posted: 03 Mar 2008 03:59 pm Post subject:

Thanks for that link Gerald!

It looks like my understanding was on the right track, but I failed to consider multiples of 5 and 10 (288 and 144). I went straight to considering numbers that represented halves 1/2, 1/4, 1/8, and 1/16.

Unfortunately the person who created the PDF files in my original link didn't think to include them either. It is odd that a multiple of 5 would be where he found his best resolution with a 4x24 cartridge.

I'm wondering if the banding effect has something to do with overlap of the cartridge nozzles or even a flaw in his printer.

I'm in the process of getting his images to print and look at. Unfortunately they aren't the same as those in PDF test sheet that I posted in my first post.

--------------------------------------------------------------------------------
Gerald Clough
Posted: 03 Mar 2008 07:21 pm Post subject:

That's another question, and I don't know enough about the printer hardware to know if there's any loss of precision as it ages and wears or even from changes in the dot nozzles. I would expect native conflicts between pixels and head configuration to present regular banding across the image and maybe aging defects to produce isolated bands.

From what I can tell of the mechanisms, it always looks to me like they would be subject to all sorts of problems of wear and environmental crud.

--------------------------------------------------------------------------------
Steve Everist
Site Admin
Posted: 03 Mar 2008 07:35 pm Post subject:

Gerald,

I printed out the image from your link at 288 and also ran into the banding of the dots in both the 288ppi and 144ppi grids.

I then converted it (no resampling) to 360ppi and the one that was considered 144ppi (assuming it converted to 180) looked better than the original, and there was no banding evident. The dots from the higher resolution grid (288 - 360) aren't as clear, although there is no banding as there was when printed out in the 288ppi configuration.

The text was clearer at 360 than at 288 also. with the 288ppi image showing a yellow fringing that wasn't visible in the 360ppi version. As far as the photograph portion of the image, I'm not seeing anything that sends me really one way or another. Since it wasn't resampled, the physical size at 360ppi has been reduced. This does appear to give it the perception of increased sharpness, but it may be mostly a difference in the physical size.

--------------------------------------------------------------------------------
George Reis
Posted: 06 Mar 2008 06:59 pm Post subject:

Your initial post on this topic, Steve, really has two parts to it - the scanning resolution; and the printing resolution.

I have never been a proponent of the 1000 PPI "standard" and find many reasons to consider other resolutions. One issue is that many scanners have to resample to scan at 1000 PPI - scanners generally don't have a continually variable resolution, so when setting a scanning rate that is not an even factor of the maximum optical resolution, there is a good possibility that the scanner is resampling the image. There are other issues involved, such as then corresponding the scanning resolution with the printing resolution, and an implication that sub-1000 PPI settings are somehow unusable, etc.

Another issue is the whole concept of printing latents 1:1. I have never had to do this in my career. I began using digital imaging with latent prints in 1992, and I've spent a lot of time helping agencies do this, but still haven't seen anyone with a good reason for doing so. Potentially, the original latent has the most information. Then, the digital file is next. The printed image is never going to equal the image quality on one's screen. And, why would want a tiny, little print anyway?

But, your main topic is printing, and it's great to see your post on this topic. It is unfortunate that printing is such a complex issue. I've seen so much bad information out there that it's hard to sort out the good stuff. Then, some police IT departments won't load the proper printing profiles for printers, so the analysts who do the printing are per-ordained to get bad images.

Anyway, good discussion. I tested my printer when I got it several years ago and find that I get excellent quality with good sharpness at 240 PPI, and a little more sharpness at the cost of color fidelity at 360 PPI (this is on my Epson 1280 - yes, I am overdue for an upgrade - I plan to convert this to a grayscale printer using four shades of black/gray inks eventually).

George

--------------------------------------------------------------------------------
Steve Everist
Site Admin
Posted: 06 Mar 2008 07:14 pm Post subject:

George,

I'm glad you weighed in on this topic. I was just about to email you regarding another issue relative to scanning lift cards that have been processed with silver powder.

I agree with your thoughts about the 1000ppi issue. I usually scan at 2400ppi as it's the highest native resolution of our scanners. It's not always needed, but I'd rather have more and need less than vice versa. The Hayden case, which most digital imaging courses reference, was much less than 1000ppi.

For some reason, after the SWG's included 1000ppi in their guidelines, it became a target to reach instead of a sort of theoretical safety net. I even remember being taught about how to get your camera/lens to capture an image as close to 1000ppi as possible, and then being told to mark on the lens that spot so that all photos could be captured as close to 1000ppi as possible. I'm glad the current trend is to just train to 'fill the frame.'

--------------------------------------------------------------------------------
Charles Parker
Posted: 06 Mar 2008 09:22 pm Post subject:

Great discussion. This may solve the problem with one of our printers putting out crap.

In discussing this and other issues today, we were told that if you scan a latent print in it is best to put it in the center of the scanning glass instead of the corner. And if you are having problems with a light print that you cover it with a gray sheet instead of using the normal white backing of the scanner lid.

I can rember using that concept on some photography years ago but it has slipped from my mind.

I have to try out everyones suggestions.

I like this kind of thread----keep it up.

--------------------------------------------------------------------------------
Mike French
Posted: 07 Mar 2008 07:27 am Post subject:

Here's another useful chart to calibrate your printer:

http://www.graphics.cornell.edu/~westin/misc/ISO_12233-reschart.pdf

--------------------------------------------------------------------------------
Biometric Collection, Analysis, Fusion and Transmission Technology Demonstrations
GENERAL INFORMATION - SPECIAL NOTICE

Solicitation Number: USA-SNOTE-080303-001
Posted Date: Mar 03, 2008
Original Archive Date: Jun 01, 2008
Current Archive Date: Jun 01, 2008
Naics Code: 334119 -- Other Computer Peripheral Equipment Manufacturing

Contracting Office Address
RDECOM Acquisition Center - Research Triangle Park, ATTN: AMSSB-ACR,
Research Triangle Park Contracting Division,
P.O. Box 12211, Research Triangle Park, NC 27709-2211

Description

Military applications of biometric exploitation to date have largely focused on the use of fingerprints to identify personnel for entry into controlled areas. Biometrics has an expanded role in military operations such as combat identification (friend, foe, or neutral); offensive operations (intelligence support to targeting); force protection (base access); detention operations; civil-military operations (track target members of a population); and personnel recovery and identification. In addition, biometrics can be utilized to improve business functions such as physical and logical access controls and privilege management applications like health care benefits, finance, and time reporting.

U.S. Army Research Laboratory (ARL) is currently seeking proposals through its Broad Agency Announcement W911NF-07-R-0001, Topic 2.63, added via Amendment 1, found at http://www.arl.army.mil/www/DownloadedInternetPages/CurrentPages/DoingBusinesswithARL/research/07_r_0001_amendment1.pdf, to develop and transition new methodologies, tools, technologies and techniques that advance the state-of-the-art across the spectrum of the biometrics infrastructure to provide users with new and enhanced capabilities. ARL solicits proposals focusing on the development and demonstration of functional models and the assessment of biometric functionality across any of the Focus Areas described below. Demonstrations and assessments will be performed in coordination with the Department of Defense (DoD) Biometrics Task Force. Results should include technical performance measurements and analysis from actual testbeds and laboratory environments.

Research Concentration Areas: In order to address these challenges, the U.S. Army Research Laboratory (ARL), on behalf of the DoD Biometrics Task Force, is soliciting proposals in the following areas. These are suggestive of topic areas that will enable future DoD systems to move toward full spectrum, biometric capabilities. They are not in priority order and are meant to stimulate ideas for perspective proposals. Other novel and creative solutions are strongly encouraged.

a. Ability to sense and collect biometric
b. Ability to enhance biometric data
c. Application and processing algorithms
d. Biometrics and all sources data fusion
e. Biometrics database and communications architectures

Proposals must show, and will be expected to deliver, a functional model package that will include usable hardware/software methodologies, models archetypes, and/or tools that can be tested and incorporated into DoD R&D programs. An open framework environment will be provided by ARL where tools and products may be deployed and allowed to interact in real-time, interactive, evolutionary and interdependent means, allowing rigorous testing of new technologies, methodologies and theories in support of the DoD Biometrics Task Force.

While Topic 2.63 will remain open throughout the period covered by BAA W911NF-07-R-0001, funding is currently available under this topic for up to ten procurement contract awards. The funding level for these awards is not anticipated to exceed $750,000/year per award. To be considered for this year's funding, white papers in accordance with PART III - SECTIONS 3-4 of the ARL BAA must be received by 21 March 2008. Each white paper must be submitted with a single PowerPoint chart, in landscape mode, divided into four equal quadrants. The upper left quadrant shall contain a graphic representing the approach or product described in the white paper. The lower left quadrant shall contain a brief description of the approach and the expected product described in the white paper. The upper right quadrant shall contain a brief description of the approach and the expected product described in the white paper. The lower right quadrant shall contain an estimate of the proposed cost for each major milestone described in the white paper. Use no smaller than 18 point Arial font. This chart will not count against the white paper page count.

In approximately 2-3 weeks after receipt of white papers, offerors submitting proposals of interest to ARL will receive an invitation to submit a complete proposal in accordance with PART III - SECTIONS 5-6 of the ARL BAA. Complete proposals will be due approximately 3-4 weeks from issuance of the invitation letter.

The proposed schedule for initial proposals/awards is:

White papers due: 21 March 20 08
Proposal Invitations: 15 April 2008
Proposals Due: 9 May 2008
Awards: 10 July 2008

Demonstrations and assessments will be performed in coordination with the DoD Biometrics Task Force. Results are to include technical performance measurements and analysis from actual testbeds and laboratory environments that indicate the effectiveness of the technology to advance the state-of-the art in current biometrics capabilities across and/or within the five (5) Focus Areas.

a. Collection Systems: The ability to sense and collect biometric data is critical to the overall biometrics process. The Collection Systems Focus Area includes the capability to obtain biometric data as well as all related biographical and/or situational data. Below are some examples of possible collection system challenges that could be addressed:

• Multi-modal biometrics data capture from both cooperative/noncooperative subjects
• Lighter, portable/handheld, tactically robust sensor systems
• Day/night and/or active/passive acquisition of biometrics data
• Ease of use/Human Factors Design

b. Sub-Optimal Data Enhancement: An increasing amount of biometric information is collected in less-than-ideal conditions, either due to the environment, mission, or the subject. These conditions are created due to limitations in collection systems or faults with the original sample and can lead to poor data quality which may limit the sample’s usefulness with current storage and search capabilities. The ability to enhance this sub-optimal biometric data will provide users with more flexibility in the collection and improved identity analysis of biometric data. Below are examples of possible sub-optimal data enhancement challenges that could be addressed:

• Signal and image processing of sub-optimal data sets from both cooperative/non-cooperative subjects
• Expert systems (calibration for unsupervised processing)

c. Next Generation Algorithms: Algorithms play a critical role throughout the biometric process. As the environment changes and new requirements are
identified, the algorithms, which carry out the computations throughout the biometrics process, must be updated to match the need. In addition, several
algorithms currently in use are proprietary, making it difficult to compare performance across various approaches. The intent for next generation algorithms is to meet the changing need and develop a product that can be shared across the DoD and the federal government. Below are examples of possible next generation algorithm challenges that could be addressed:

• Detection, segmentation, matching, and quality
• Adaptive algorithms focused on newly developed sensors and sensor modalities
• Post-capture processing algorithms for existing data (non-cooperative)
• DoD owned/developed algorithms for existing data (e.g., iris, fingerprint, facial)
• Robust algorithms three-dimensional modeling to improve face recognition for difficult images and techniques to accommodate human attributes (e.g. face aging, weight gain/loss)
• Algorithms that execute on many processor architectures, embedded/small devices

d. Biometrics Data Fusion: Traditional biometric systems focus on the use of a single biometric modality (e.g., fingerprint) for the verification and identification of individuals. However, relying on only a single modality limits the matching accuracy and the capability of analysts. Future biometric technologies must be able to combine multiple biometric modalities and other non-biometric data to allow for improved matching and more robust identity analysis. Below are examples of possible biometrics data fusion challenges that could be addressed:

• Feature, score, decision-level algorithms for data fusion
• Mission-specific data presentation
• Contextual data fusion with biometric data
• Standardization of training samples and data formats
• Tagging, Tracking, and Locating link analysis (information or entity level fusion)

e. Biometrics Architecture: The use of biometrics as an enabling technology requires the support of a complex architecture that must be updated continuously to meet emerging requirements. This focus area includes those topics not addressed in the previous focus areas. Some of the topics in this focus area are addressed by providing new approaches or methods to assess performance rather than a technology solution. Below are examples of possible biometrics architecture challenges that could be addressed:

• Processor/communications and data storage trade-off analysis
• Modeling and simulation
• Scalability
• Security architecture
• Movement and management of data

http://www.arl.army.mil/www/DownloadedInternetPages/CurrentPages/DoingBusinesswithARL/research/07_r_0001_amendment1.pdf

Point of Contact
Dr. Keith Aliberti, keith.aliberti@us.army.mil, (301)-394-2320.

_________________________________________

KEPT - Keeping Examiners Prepared for Testimony - #10
Analysis phase of ACE-V
by Michele Triplett, King County Sheriff's Office

Disclaimer:  The intent of this is to provide thought provoking discussion.  No claims of accuracy exist. 

Question – Analysis Phase:

Can you explain the Analysis phase of ACE-V?

Possible Answers:

a)      I’m looking for sufficiency, to see if there’s enough for me to work with.

b)      I’m looking to see how much level 1 detail, level 2 detail, and level 3 detail exists and I determine if there’s enough for me to move on to the comparison phase.

c)      I’m collecting information which includes orientation, direction, ridge flow, pattern type, ridge color, clarity (contrast level, distortional aspects).

Discussion:

The analysis phase can be described in a variety of ways.  My initial thought is that it’s not really important how you describe it, but the truth is that if several people are testifying in the same case (which is happening more and more every day) and they describe anything differently then it appears as though there’s not a uniform method.  The difference in explaining something could also make it appear that one of the practitioners isn’t as knowledgeable as they should be.  To insure this isn’t the interpretation left in court we do need to have similar explanations.

Answer a:  This isn’t a bad answer but it really doesn’t describe the analysis stage.  I think it describes a pre-analysis stage or a quick assessment of what may be on a lift card.

Answer b:  This isn’t a bad answer either but it brings in a lot of terms that the court may not be familiar with.  Using an answer like this may result in you having to do a lot more explaining than is necessary.

Answer c:  This is a simple answer that can easily fit into most agencies use of the analysis phase.  It states ‘what’ you’re doing (collecting information) without stating ‘how’ it should be done.  This would fit agencies that print pictures of a latent print and mark the level 2 details and it would also fit agencies that just do a visual inspection.  This conclusion also conforms to the scientific method of hypothesis testing (where the first stage is to collect data).

_________________________________________

Feel free to pass The Detail along to other examiners.  This is a free newsletter FOR latent print examiners, BY latent print examiners. With the exception of weeks such as this week, there are no copyrights on The Detail content.  As always, the website is open for all to visit!

If you have not yet signed up to receive the Weekly Detail in YOUR e-mail inbox, go ahead and join the list now so you don't miss out!  (To join this free e-mail newsletter, enter your name and e-mail address on the following page: http://www.clpex.com/Subscribe.htm  You will be sent a Confirmation e-mail... just click on the link in that e-mail, or paste it into an Internet Explorer address bar, and you are signed up!)  If you have problems receiving the Detail from a work e-mail address, there have been past issues with department e-mail filters considering the Detail as potential unsolicited e-mail.  Try subscribing from a home e-mail address or contact your IT department to allow e-mails from Topica.  Members may unsubscribe at any time.  If you have difficulties with the sign-up process or have been inadvertently removed from the list, e-mail me personally at kaseywertheim@aol.com and I will try to work things out.

Until next Monday morning, don't work too hard or too little.

Have a GREAT week!