|
G o o d
M o r n i n g !
via
THE WEEKLY DETAIL
Monday">
|
_________________________________________ |
__________________________________________
Breaking NEWz you can UzE... |
|
compiled by Jon Stimac |
|
Suspect Says Ankle Monitor Proves He's Not Burglar
–
PIONEER PRESS, MN
- Jan 5, 2008
- ...police are trying to figure out how the suspects fingerprints
allegedly turned up at the scene...
Broken Fingerprint Machine Led to Mistaken Inmate Release, Union Says
–
MILWAUKEE JOURNAL SENTINEL, WI
- Jan 3, 2008 -
...machine located in the jail's release area has been out of order
for about a year...
Fingerprints, DNA Analysis Help Police Develop Leads –
ERIE TIMES, PA - Jan 2, 2008
...the lab entered the print into the FBI's IAFIS, which contains more
than 47 million prints...
CSI Team is Real; the Crime's Fake
–
FLORIDA TIMES-UNION, FL
- Dec 28, 2007
...students
learn how to collect fingerprints and study blood, hair and fibers
from fake crime scenes... |
__________________________________________
Recent CLPEX Posting Activity |
Last Week's
Board topics
containing new posts
Moderated by Steve Everist |
|
Calls for Inquiry to be scrapped
Daktari 4705 Sun Jan 06, 2008 8:04 pm
Evidence Fabrication in South Africa
Pat A. Wertheim 4320 Sun Jan 06, 2008 1:44 pm
Technology Bah-Humbug
Charles Parker 436 Sat Jan 05, 2008 11:58 pm
Watchdog groups urge ‘CSI’ toy recalls
RedFive 208 Fri Jan 04, 2008 7:10 pm
Need published studies on results of casing processing...
Amy Miller 198 Fri Jan 04, 2008 7:04 pm
Zero Error Rate Methodology
Charles Parker 583 Fri Jan 04, 2008 2:45 am
Zero Error Rate vs. No Error Rate
Michele 377 Thu Jan 03, 2008 7:52 pm
DNA and Fingerprints
Charles Parker 622 Mon Dec 31, 2007 9:23 pm
Teaching the kids
charlton97 192 Sun Dec 30, 2007 11:44 am
(http://clpex.com/phpBB/viewforum.php?f=2)
|
UPDATES ON CLPEX.com
Updated the Fingerprint Interest Group web page with FIG #
27.
Updated the Detail Archives
_________________________________________
we had a happy holidays and hopefully a good
catch-up week as we swing into 2008.
we start a new regular
column from Michele Triplett entitled KEPT - Keeping Examiners Prepared for
Testimony. Make sure to check out the end of each Weekly Detail for
this regular column, and thanks to Michele for providing this important
topic!
This week our main feature is the NIST CONOPS involving the upcoming latent
print technology evaluation:
_________________________________________
Concept of Operations (CONOPS) for Evaluation of Latent Fingerprint
Technologies (ELFT)
by NIST
http://fingerprint.nist.gov/latent/elft07/elft_p2_concept.pdf
(Rev. E, 1 Nov 2007)
1. Overview
The National Institute of Standards and Technology (NIST) is conducting a
series of tests for evaluating the state of the art in automated latent
fingerprint matching. The intent of the testing is to quantify the core
algorithmic capability of contemporary matchers. The testing will be
conducted using software-only implementations, and utilizing NIST hardware.
The umbrella project for the series of tests has been named Evaluation of
Latent Fingerprint Technologies (ELFT). The scope and structure of these
tests are based partly upon lessons learned from the April 2006 NIST Latent
Fingerprint Testing Workshop, supplemented by technical interchanges with
workshop participants and vendors. The initial round of tests was initiated
in April 2007. Computer runs concluded in July, and the analysis of the
results was posted on the website in September 2007.
The principal objective of this report is to provide a “snapshot” of the
thinking, analysis, and planning that went into ELFT. The report is not
intended to provide any test results or conclusions. These will be presented
in subsequent reports. While the immediate goal of ELFT is to assess
automated technology, the long-term goals go far beyond simply quantifying
performance. It is fully expected that understanding the performance
envelope and limitations of contemporary matchers will lead to improvements
in technology. These in turn, will lead to enhanced performance for searches
of ten-prints and plain impressions against unsolved latent databases/watchlists.
Equally important, technology improvements will provide law enforcement the
capability to search their unsolved latent fingerprints against ten-print
files with greatly reduced effort.
ELFT is structured as a multi-year project, and the full impact of this work
may not be felt for several years. The first part of this project, ELFT07
(07 is the year), consists of two tests, run in a “lights-out” environment.
The two tests have been termed Phase I and II. Phase I is a proof of concept
test, whose main purpose is to demonstrate integrity of the software in a
lights-out environment. During Phase I the software will demonstrate: a)
automated feature extraction from latent images; b) the ability to match
these features against enrolled 10-print backgrounds; and c) generation of
candidate lists. Phase II will then employ a larger database to quantify the
achievable performance (“hit rate”) for automated searches.
In subsequent years (2008+) we plan to expand the above tests in several
ways. First, we plan to augment the ten-print databases with a mix of rolled
and plain impressions (“flats”). These will enhance NIST’s and the latent
community’s understanding of the challenges of matching latents against
flats. Continuing this line of investigation, we will then transition to
searches of plain impressions against databases of latent images (sometimes
referred to as reverse searches). Initially these tests will be restricted
to single-finger searches, and subsequently will be enlarged to multi-finger
latent searches.
Envisioned tests include: a) latents and mates scanned at enhanced
resolution (1000 and 2000 ppi); b) latents lifted/developed/processed in
diverse manner (i.e., how was the image actually produced from the latent);
c) latents matched against latents; and d) searches employing new or
non-traditional features (e.g., level-3 features).
NIST is also looking into the development of latent image quality measures (LIQM).
The principal function of the latent image quality measure is to provide a
good indication of whether a latent is amenable to automated (“lights-out”)
matching. Only latent prints of
higher quality measure would be submitted for lights-out matching. The
development of a suitable quality measure might require additional testing.
We have outlined above the full scope of this project with a “broad brush.”
In the remainder of this document we focus on this year’s portion of the
project, ELFT07. So as not to break up the flow, more detailed discussion of
select topics have been moved to the end, Section 14.
2. Who Should Participate in the Tests?
Developers of latent fingerprint matcher software systems are strongly
encouraged to participate in ELFT07. In addition, companies, research
organizations, or universities that have developed mature prototypes, or
stable research latent fingerprint matchers, are invited to participate. The
latent fingerprint matching software submitted need not be a “production”
system, nor be commercially available.
3. Precedence of Documents
It is intended that this Concept of Operations (CONOPS or ConOps) be the
single most comprehensive document covering Latent Testing/ELFT concepts. It
will be periodically revised to reflect updates in NIST’s planning, and to
incorporate vendor comments and suggestions. However, this CONOPS is not
guaranteed to be the most accurate for highly technical data. In the event
of conflict with the Application Programming Interface (API) or the
Application Form the latter two will take precedence. These two documents
may be supplemented by additional reference documents in the future.
4. Test Objectives – What will be Tested during ELFT07?
As previously indicated, the primary purpose of the Phase I and II testing
is to quantify the core algorithmic capability of contemporary matchers, in
order to understand their strengths and limitations. In the initial tests
the emphasis is on matching latents against ten-prints. (In subsequent tests
this will be expanded to include other types of matching, for example
latents against plain-impressions, and latents against latents.) The testing
will be conducted using a software-only implementation, in a lights-out
environment, and utilizing NIST hardware and datasets. During Phase I the
software will demonstrate: a) automated feature extraction from latent
images; b) the ability to match these features against enrolled 10-print
backgrounds; and c) generation of candidate lists.
While Phase I is primarily intended to be a proof of concept test, it will
nevertheless provide a certain amount of performance statistics. Only
aggregate statistics (combining all successful participants) will be
published. Individual results will be disclosed only to the owner of the
respective SDK.
NIST will compute (and report as an aggregate statistic) the number of
“hits” in each position (first, second, third, up to 50).
Figure 5 – Specimen Cumulative Distribution
NIST will also report two overall performance metrics:
1) Metric #1 simply counts the fraction of cases in which the correct mate
appears
in top position on the candidate list. This metric ignores all candidates
below
top position.
2) Metric #2 gives partial credit for mates appearing in lower than top
position. It
assign 1.0 to a mate in top position, 0.5 to one in second position; 0.3333
to
third position, etc. The final score is the sum of these scores divided by
the
number of searches with mates. (For the validation data all searches have
mates.) This score will always be at least as high as metric #1, and will
generally be somewhat higher. The highest possible score is 1.0 and the
lowest
is 0.
NIST will also use DET performance metrics as a primary indicator of
one-to-many
identification search accuracy. This involves plotting False Acceptance Rate
(FAR) and
True Acceptance Rate (TAR) for all values of the threshold. (Equivalently,
one can use
false rejection and false acceptance rates.)
For Phase II we will employ a larger database so as to improve the
statistics. The results will be published as aggregate statistics, as well
as individual participant statistics (for those participants electing to
continue to Phase II). NIST will also report enrollment and search timing
information. Speed of execution, for both enrollment and latent search, is
of secondary importance. However, in order to conduct these tests in a
reasonable amount of time NIST must impose some limitations. These are
covered in Section 11. In reporting timing measurements NIST will specify
the exact hardware that the software was hosted on. NIST will in addition
caveat timing measurements by noting that operational latent searching
algorithms are likely to be implemented in more sophisticated hardware.
NIST recognizes that latent searches pose many special problems. One of
these is that “strong hits” may have widely different matcher scores (for
example, depending upon the number of minutiae). This may place challenges
on the DET approach. Largely for this reason we have recommended the
inclusion of a normalized score that attempts to compensate for the matching
score variations. See Sections 13 and 14 for further discussion.
A comprehensive discussion of performance metrics and their definitions is
found Patrick Grother, Ross Micheals, and P. Jonathon Phillips, Face
Recognition Vendor Test 2002 Performance Metrics, 31 March 2003. See also
http://ois.nist.gov/nistpubs/technipubs/recent/search.cfm?dbibid=13808
5. Publication of Participation and Results
NIST understands that this project is entering a relatively unexplored
field, and many challenges lie ahead. For this reason we have structured
ELFT07 to include two phases. We consider Phase I to be a Proof of Concept
Test. This means that the primary objective of Phase I will be to
demonstrate that the submitted SDK executes on the Phase I data to
completion, in a lights-out environment, and produces a “meaningful” output.
A “meaningful” output is basically an output in the correct format.
The detailed results of Phase I will be discussed with the Participant on a
“one-on-one basis,” but will not be published or submitted to other
government agencies. The number, but not the names, of participants who
attempted and completed Phase I will be disclosed. However, in the (likely)
event there are a significant number of participants in Phase I, NIST is
considering publishing the aggregate test results (under the premise that
this may be a fair assessment of the state of the art). By “aggregate test
results” we mean that results are “lumped,” and that no specific candidate
list or participant-specific scores will be mentioned. Participants will
have the option to withdraw anonymously following participation in Phase I.
(This means that their withdrawal will not become a public announcement.)
Participants who elect to continue to Phase II may resubmit their SDKs.
These need not be identical to those of Phase I. Following completion of
Phase II testing the Government will combine all results into a Final
Report. The Evaluation of Latent Fingerprint Technologies Test, Phase II
Final Report will contain, at a minimum, descriptive information concerning
ELFT07, descriptions of each experiment, and aggregate test results. Should
individual participant’s scores be published, NIST will exercise care that
any implied rankings are well supported by the underlying statistics. (That
is, two scores should be considered essentially the same if the difference
is significantly less than the error bar.)
Participants will have an opportunity to review and comment on the Final
Report. Participants’ comments will be either incorporated into the main
body of the report (if it is decided NIST reported in error) or published as
an addendum. Comments will be attributed to the participant. After the
release of the Phase II Final Report, Participants may decide to use the
results for their own purposes. Such results shall be accompanied by the
following phrase: “Results shown from the Evaluation of Latent Fingerprint
Technologies Test (ELFT07) do not constitute endorsement of any particular
system by the U. S. Government.” Such results shall also be accompanied by
the Internet address (URL) of the ELFT07 Final Report on the ELFT07 website.
For Phase III and beyond NIST intends to publish statements of the
performance of all implementations submitted for testing. These will include
measurements of identification error rates and throughput. These results
will be attributed to participants. Accordingly, NIST will require an
appropriately signed application form from all participants and NIST will
not evaluate any implementation unless the participant consents to the
disclosure of its performance. The NIST tests use sequestered images. These
will not be provided to participants.
6. Protection of Participant’s Software
NIST recognizes the proprietary nature of the participant’s software and
will take all reasonable steps to protect this.
7. Why “Lights-out”?
The term “lights-out,” as used in this document, will indicate that no human
assistance will be required in conducting the latent searches. In
particular, all feature extraction steps, both for the enrolled images and
for the latent images, must be performed entirely by the SDK under test.
There are good reasons why NIST selected the “lights-out” mode of testing:
1. It decouples the skill of the human expert from the intrinsic merits of
the software.
2. It protects the privacy of the test data by keeping the data in house,
and not requiring examination by non-government personnel. This mode of
testing allows the use of Sensitive but Unclassified test data.
3. It encourages a forward-looking view of how latent searches might be done
in the near future. It is anticipated this broader outlook will lead to
technical innovations. Algorithmically speaking, “lights out” consists of
two separate concepts. The first is automated feature extraction (and of
course matching). The second is candidate list reduction. NIST envisages
that in the near future automated search capabilities will assist latent
experts by reducing the size of candidate lists that they need to examine by
eliminating the more obvious “nuisance” non-matches. We refer to this part
of the automated matching as candidate list reduction. To achieve effective
candidate list reduction may require additional computer processing,
including the development of new algorithms.
Although this is a “lights-out” test NIST will use some human assistance in
the data preparation phase. Any such assistance will be provided indirectly
by NIST, and might include a) cropping and/or re-orienting of selected
latent images, and b) specifying a region-of-interest in the from of a mask.
The mask will be a byte image conformal with the size of the latent image.
Initially only two values will be used for each pixel, 0 and 255. A zero
value will indicate “do not use this pixel,” while 255 will indicate a
“good”
pixel. In the future these two values may be augmented by other values to
indicate finer gradations of quality. NIST will also involve latent experts
for examining potential consolidations and for resolving contested or
unclear results.
8. Test Data
NIST will select the test datasets from its internal sources. The Test
Datasets are protected under the Privacy Act (5 U.S.C. 552a), and will be
treated as Sensitive but Unclassified and/or Law Enforcement Sensitive.
ELFT07 Participants will have no access to ELFT07 Test Data, neither before,
during, or after the test, with the exception of the small Validation
Dataset.
8.1 ELFT07 Datasets
The Validation Dataset is a very small dataset intended to demonstrate that
the received software (SDK) is stable and compliant with the API. Upon
receiving the applicant’s SDK and Validation Dataset results, NIST will
rerun the applicant’s software using the Validation Dataset. For the
applicant to be officially accepted (and designated a Participant) NIST must
be able to reproduce the submitted results.
For Phase I the Validation Dataset consisted of ten latent searches and 100
background ten-prints. For Phase II this will increase somewhat, say 15
searches. Problems encountered during Phase I demonstrated the need to
include test cases for “exceptions,” for example missing fingers. In the
event of disagreement in the two outputs, or other difficulties, the
Participant will be notified. Participants will be notified with a detailed
description of the problem(s) and given reasonable opportunity to resubmit.
Both Phase I and Phase II Test Datasets consist of latent images for
searches and tenprints for the background (“gallery”). The Phase I
Validation dataset was very “benign,” and avoided known problem cases such
as: 1) very small latent area; 2) extremely busy or otherwise difficult
background; 3) highly blurred image; 4) multiple fingerprint impressions; or
5) upside down or mirror images. The (main) Phase I Dataset was similar,
though it did include a small number of “problem cases.” The Phase II
dataset will be principally drawn from casework, but will for the most part
avoid the more difficult cases.
Latent images will be supplied uncompressed, and will have been scanned at
either 500 ppi or 1000 ppi. The participant should be prepared to handle
either resolution. Additional image characteristics may be found in the API.
Background (“gallery”) data will consist of rolled ten-print impressions,
scanned at 500 ppi, and presented in a decompressed form.
8.2 Size of Images
For ELFT07 the following guidelines apply:
v All mates (background) will be rolled impressions. There will be no
“flats.”
v In all cases rolled impressions will not exceed 1000 x 1000
v In all cases latent images will not exceed 2000 x 2000, though often will
be significantly smaller.
v Minimum dimensions for latent images under 300 are possible, but never
smaller than 150.
Over the entire series of planned tests the size of test images may vary
considerably. For example, images scanned at 2000 ppi will contribute some
very large sizes, potentially as large as 4000 x 4000 pixels. However, these
very large sizes will only appear in the “downstream” tests.
9. Testing Platform
NIST will host the participant’s software (SDK) on a high-end PC
(workstation/server type). Although these PCs include of a mix of models, a
“typical” PC will have the equivalent of a Pentium 4, 2.8 GHz processor, or
higher; 2 GB of memory; and at least 50 GB of disk memory. The participant
software must be able to reside and execute on this single PC. NIST, at its
discretion, must be able to copy the software to several PCs to expedite or
scale-up the testing. These computers are configured with either a Windows
2000 or Linux operating system.
10. Format of Participant Software
The software undergoing testing will be hosted on NIST-supplied computers.
The executable modules will be built up from two sources: 1)
participant-supplied software provided in the form of a Software Development
Kit (SDK), and 2) NIST-supplied software. The core of the executable module
is of course derived from the SDK. The part supplied by NIST is mainly
concerned with the image retrieval and manipulation.
13. Format of Candidate List
The output candidate list should have a fixed length of fifty (50)
candidates. We have selected this size because it is short enough to be
convenient, yet long enough to give an indication of the number of “hits
just out of reach.” (We currently don’t envision cases in which the
background is less than 50 fingers. Should this situation arise, the
candidate list could be suitably “padded.”) The candidate list consists of
two parts, a required and an optional part. The required part consists of:
1) the index of the mating ten-print subject; 2) the matching finger number;
3) the absolute matching score; and 4) an estimate of the probability of a
match (0 to 100, see also Section 14). The optional part consists of: 5) the
number of good minutiae identified in the latent; 6) the number of latent
minutiae which were successfully matched; 7) the quality estimate of the
latent (0 to 100, 100 is best); and 8) the quality estimate of the mate (0
to 100, 100 is best). The API provides further guidelines regarding the
meaning of quality scores. The candidate list is ordered based upon the
absolute score, highest score in first position.
14. Supplemental Notes
14.1 Supplemental Notes to Section 1.0 -- Concepts of Operation for
“Improved Watchlist Searches”
A major goal of this project is to improve searches of watchlist/lookout-lists.
It is becoming increasingly common to capture live fingerprints of arriving
passengers at ports of entry and similar venues. The capture fingerprints
are then compared to fingerprints of the person on file (one-to-one
comparison, or validation match); they may also be searched against selected
watchlists (one-to-many search).
The newly acquired fingerprints may then be matched against any or all of
three types of finger prints. Verification, or one-to-one searches, are
generally performed by matching with plain impression taken at a previous
time. Watchlists or Lookout-lists contain the fingerprints of prior
offenders, or persons of interest. They may be comprised of any of the types
of fingerprints, though latent fingerprints are generallykept in separate
files, exclusively dedicated to latents.
Plain-impression may be matched against any of three types of fingerprints.
The simplest and clearest example of the applicability ELFT is to watchlists
comprised of latents. However, there are several other ways in which ELFT
may contribute to the point-of-entry scenario:
Ø Low-quality livescan images provide many of the same challenges as do
latents. Improvements in latent matching should therefore transfer to
real-time livescan matching. (Livescan images they are subject to “retake,”
but the number of retakes is necessarily very limited, because of the need
to expedite the processing.)
Ø To provide searches of watchlists in near-real-time, substantial
algorithmic improvements are required. The multi-stage matching approach
used by some latent matchers may offer a solution.
Ø For increased search accuracy, additional features (e.g., level 3) might
be required. A goal of ELFT to examine the performance increases provided by
selected new features.
14.2 Supplemental Notes to Section 1.0 -- Concepts of Operation for
“Improved Criminal Latent Searches”
A second major goal of ELFT is to provide “an automated latent search
capability” to latent examiners. By this we mean that latent examiners
should have the capability of screening their latent images with a minimum
of effort. We use the term screening to emphasize that such searches are not
fully equivalent to traditional searches.
14.3 Features for Use in Matching
Generally speaking, the selection of the features for use in the matching
process is left to the participant. Matchers need not primarily be a
“minutiae matcher.”
Candidate List architectures in which “advanced matchers” are selectively
invoked depending upon initial results are allowed. For example the matcher
might initially use certain core features in comparing the search (probe)
with a background (gallery) subject. The result of this comparison might
produce one of three possible outcomes: 1) the two fingerprints are too
different, and no further effort should be expended on this candidate; 2)
the two are so similar that this is definitely a mate; or 3) the two have
points of similarity, but the match is not conclusive. The third case might
then trigger a call to an “advanced matcher” for further resolution.
If using “advanced matchers”, it is up to the participant to decide if the
additional features (if any) required are to be extracted and stored on disk
memory during the enrollment phase. Since it may not be possible to keep all
gallery images in memory, it might be necessary for the software to retrieve
the data from disk during searches. This extra fetch time will be included
in execution time measurements.
Approximately a year-and-a-half downstream NIST intends to test the effect
of using augmented feature sets. These will be largely based upon the CDEFFS
feature sets, but are not necessarily limited to these. For a description of
the proposed CDEFFS features please go to the following website: http://fingerprint.nist.gov/standard/cdeffs/index.html
(http://fingerprint.nist.gov/standard/cdeffs/index.html)
These augmented feature tests test will be run in a dual mode: a) first
without employing any new feature, then b) employing designated new
features. To allow this mode of operation a somewhat different format might
be required for the SDK.
14.4 Supplemental Notes to Sections 7 and 13 -- Candidate List Reduction
NIST envisages that in the near future automated search capabilities will
assist latent experts by reducing the size of candidate lists they need to
examine through elimination of the more obvious “nuisance” non-matches
(impostors). For example, assuming that one hundred latents are submitted
for searches, and that each search produces a candidate list of twenty
candidates, an examiner needs to look at 2000 candidates. Since a typical
identification rate for latent searches might be around 4%, this means that
2000 candidates need to be examined to find the four true identifications.
While it is true that skilled examiners can quickly dismiss “nuisance
candidates,” nevertheless it does take up valuable examiner time. An even
larger concern is that too many nuisance candidates might result in the true
mates being overlooked due to operator fatigue. It is therefore desirable to
minimize these nuisance candidates. We refer to this part of the automated
matching process as candidate list reduction. To achieve effective candidate
list reduction may require additional computer processing, including the
development of new algorithms.
Since candidate list reduction poses many challenges, we plan to implement
it in stages. The initial stage is to introduce a new parameter called
Probability of True Match. This is intended to give a numerical estimate
that the candidate is a true mate of the latent. This parameter should be
supplied as a number between zero and 100. The number 100 will be
interpreted as an extremely high confidence “hit.” The intent is to use this
parameter as a key to candidate list reduction. Certainly the raw matcher
score by itself provides a strong clue regarding the merit of a given
candidate. However, by itself it is insufficient. For one thing, there is no
agreed upon standard for the range of matcher scores: Does a value of 5000
indicate a high score? A very high score? Secondly, whether a given score
belongs to a true mate depends upon the size of the background. The larger
the background the more likely it is that large impostor scores will be
created. The Probability of True Match therefore needs to take matcher score
and background size into account. Additional information might also be
factored in, such as: a) the score gap to the next candidate; b) the quality
of the latent; and c) the quality of the mate.
There does not appear to be any simple way of computing Probability of True
Match, and participants are encouraged to develop their own procedures.
Purely as an illustrative example, we offer the following procedure.
Assume a background size of N. Assume additionally that some candidate has
achieved a score of S. To fix our ideas assume S = 5000. Assume further that
data such as shown in Figure 6 is available. Then from this figure we obtain
TAR = 0.3 and FAR = .00001. This may be interpreted as: the a priori
probability of obtaining a score exceeding 5000 is 0.3 when matching against
the true mate. Conversely, the a priori probability of exceeding 5000 when
matching against an imposter is .00001. Note that this is for a single
imposter. The probability that one or more imposters exceed 5000 in a
background of N, assuming independence in match scores, is 1 – (1 – FAR)N .
If this is taken as the a priori probability of an imposter exceeding 5000,
then we can renormalize in the Bayesian sense to obtain the probability that
the candidate is a true mate.
http://fingerprint.nist.gov/latent/elft07/elft_p2_concept.pdf
_________________________________________
KEPT -
Keeping Examiners Prepared for Testimony
#1 - Comparison Phase - Documentation
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 – Comparison Phase - Documentation:
Did you document what your conclusion was based on (did
you chart it, draw it, or include a written explanation)?
Possible Answers:
a)
No.
b)
Yes, our office does this for every individualization.
c)
No, I have never done this.
d)
No, it’s not in our operating policies to do it.
e)
No, we don’t have the manpower to do this.
f)
No, it wasn’t needed in this case.
g)
No, we don’t document what we see because we don’t want to influence
others.
h)
Under ASCLD, the latent lift is considered sufficient documentation.
Discussion:
Answer a: Previously, it was recommend to
answer with a simple ‘yes’ or ‘no’ but that isn’t as acceptable as it once
was. If you just answer the above question with ‘No’ then it appears you
did a less than acceptable job….that your work was lacking.
Answer b: Perhaps some people make a chart for
every individualization. This may look like a good practice but it’s not
needed. It may look like you’re being thorough when you’re really just
wasting time by doing unnecessary steps.
Answers c, d, and e: These are excuses and it
leaves people with the impression that you should have done some form of
documentation.
Answer f: I feel the best answer is f;
documentation wasn’t done because it wasn’t needed. We all have our own
standards, agency standards, and industry standards but the most accepted
standard is the scientific standard. Science doesn’t require documentation
for simple conclusions (stating that 8x8 is 64 is good enough; I don’t have
to document each tick mark. But if I’m ever asked to produce documentation,
I should be able to). It’s also important to note that science doesn’t
require contemporaneous documentation on simple conclusions like this. When
conclusions are more complex, then documentation should go along with the
conclusion so others know the conclusion had good justification behind it.
Answer g: This may sound like a good answer but
it’s really just an excuse for not understanding or following scientific
protocols. Science requires that conclusions be documentable in case anyone
should ever ask for them.
Answer h: The first thing to notice about this
answer is that they use the term ASCLD instead of ASCLD/LAB. It’s always
more professional to use the correct name of an organization (preferably not
the acronym). This answer also seems to misunderstand the question.
Documentation of the physical evidence found is very different from
documentation of the justification behind an analytical conclusion. The
latent print itself cannot be sufficient justification for a conclusion.
_________________________________________
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!
| |
|