Hutch in the 1911 Census?

[Ben]

My scientific background tells me that we need to see empirical evidence - studies, papers (electronic or otherwise!), research - that proves that using scanned images is significantly worse than using original documents, before we believe the "just so" stories we've had so far.

But your doubt is unreasonable, with respect, Gareth, since every document examiner knows that an analysis of the original documents is necessary. The idea that they all alighted on this decision, collectively, for no good reason and without any compelling reason to do so, is regrettably outlandish.

The fact that they alll agree on this necessity is sufficient to rule out the idea that they all just "decided" on the assumption that originals were preferable to scans. The reasons offered by the expert practitioners in the field are eminently logical ones, and Crystal has outlined several of these herself, one of which included pen pressure, which is readily detectable in orginal documents in a way that cannot compare to a computerized image. Experts will of course disagree on certain aspects, but it is significant that they all agree on this issue (which is why "proof" in this case, cannot be quantified for the purposes of the "experiments" you're suggesting), and I don't consider it reasonable that they must all be considered wrong-headed in this regard.

If that was the case, it's impossible to accept that they wouldn't have been exposed by now.

ARE they, though, Ben? And, if they are, did Sue Iremonger actually use them?

It really is not good enough assuming that she did. Not good enough at all

Is is when we know that the alternative is too unutterably outlandish to contemplate.

Best regards,
Ben

[truebluedub]

Here's an abstract, I'll track down more.

Title: PROFICIENCY OF PROFESSIONAL DOCUMENT EXAMINERS IN WRITER IDENTIFICATION
Journal: Journal of Forensic Sciences Volume:39 Issue:1 DatedJanuary 1994) Pages:5-14
Author(s): M Kam ; J Wetstein ; R Conn
Sponsoring Agency: US Dept of Justice
Federal Bureau of Investigation
United States
Publication Date: 1994
Pages: 10
Type: Technical reports
Origin: United States
Language: English
Contract No.: J-FBI-91-352
Annotation: This experiment found that professional document examiners from the Federal Bureau of Investigation (FBI) were significantly better in performing writer identification than college-educated nonexperts.
Abstract: A comprehensive writer identification test was designed and administered to a group of professional document examiners and a control group of nonprofessionals. The professional group consisted of seven document examiners from the FBI, while the control group consisted of 10 graduate students in the areas of engineering and business. A data base of 86 documents created by 20 writers was used for the test. Each participant was given the same 86 documents and was instructed to sort them into separate piles. Each pile was to include documents created by the same writer. Professional document examiners from the FBI performed significantly better than control group members, indicating that handwriting identification expertise exists. The findings call into question the conclusions of previous studies in this area. Methods used by professional document examiners are discussed. 16 references and 5 tables
Main Term(s): Police effectiveness
Index Term(s): Federal Bureau of Investigation ; Document analysis ; Evidence identificatn and analysis/ ; Handwriting analysis ; Technology ; Criminal investigation/ ; Forensic sciences

To cite this abstract, use the following link:

http://www.ncjrs.gov/App/Publications/abstract.aspx?ID=146638

regards
Chris Lowe

[truebluedub]

Journal of forensic sciences:

Volume 46, Issue 4 (July 2001)




ISSN: 0022-1198
Published Online: 1 July 2001
Page Count: 5


Click here to download this paper now for $25

View License Agreement

Signature authentication by forensic document examiners
Kam, M
Data Fusion Laboratory, Electrical and Computer Engineering Department, Drexel University, Philadelphia, PA.

Gummadidala, K
Data Fusion Laboratory, Electrical and Computer Engineering Department, Drexel University, Philadelphia, PA.

Fielding, G
Data Fusion Laboratory, Electrical and Computer Engineering Department, Drexel University, Philadelphia, PA.

Conn, R
RABA Technologies, Inc., 10500 Little Patuxent Parkway, Suite 790, Columbia, MD.


Abstract
We report on the first controlled study comparing the abilities of forensic document examiners (FDEs) and laypersons in the area of signature examination. Laypersons and professional FDEs were given the same signature-authentication/simulation-detection task. They compared six known signatures generated by the same person with six unknown signatures. No a priori knowledge of the distribution of genuine and nongenuine signatures in the unknown signature set was available to test-takers. Three different monetary incentive schemes were implemented to motivate the laypersons.
We provide two major findings: (i) the data provided by FDEs and by laypersons in our tests were significantly different (namely, the hypothesis that there is no difference between the assessments provided by FDEs and laypersons about genuineness and nongenuineness of signatures was rejected); and (ii) the error rates exhibited by the FDEs were much smaller than those of the laypersons. In addition, we found no statistically significant differences between the data sets obtained from laypersons who received different monetary incentives.
The most pronounced differences in error rates appeared when nongenuine signatures were declared authentic (Type I error) and when authentic signatures were declared nongenuine (Type II error). Type I error was made by FDEs in 0.49% of the cases, but laypersons made it in 6.47% of the cases. Type II error was made by FDEs in 7.05% of the cases, but laypersons made it in 26.1% of the cases.

Keywords:
examination, forensic science, handwriting, questioned documents, signatures, validation

regards
Chris Lowe

[truebluedub]

Citation Information
Journal of forensic science
Stock #: JFS9511
Volume: 40
Issue: 6
Year: 1995
Pages: 1045-1051
Author(s): Hicks AF
Title: Computer Imaging for Questioned Document Examiners I: The Benefits
Keywords: computer imaging, computers, digital image processing, document examiners, forensic science, questioned documents


Abstract: Computers are touching every aspect of our lives. Many computer software programs are beneficial to Forensic Document Examiners. These include word processing programs and databases for typewriter reference files. A new area of interest is the use of digital imaging programs to address some document problems. Many imaging packages, such as Adobe Photoshop 3.0, allow the examiner to quickly and easily tackle problems that, otherwise, would require some very expensive instrumentation and/or a great deal of time and trouble. This software can also be of benefit in preparing demonstrative exhibits of the examiner's findings. Illustrations and step-by-step instructions will assist examiners in putting this technology to work for them.

kind regards
Chris Lowe

[truebluedub]

JFS Volume 47, Issue 4 (July 2002)




ISSN: 0022-1198
Published Online: 1 July 2002
Page Count: 17


Click here to download this paper now for $25

View License Agreement

Individuality of handwriting
Srihari, SN
University distuinguished professor, Department of Computer Science and Engineering and Director, Center of Excellence for Document Analysis and Recognition, University at Buffalo, State University of New York, Buffalo, NY 14228.

Cha, SH
Assistant professor, Pace University, Pleasantville, NY 10570.

Arora, H
Research scientist, IBM, Endicott, NY.

Lee, S
Doctoral candidate, Department of Computer Science and Engineering, Unversity at Buffalo, State Unversity of New York, Buffalo, NY 14228.


Abstract
Motivated by several rulings in United States courts concerning expert testimony in general, and handwriting testimony in particular, we undertook a study to objectively validate the hypothesis that handwriting is individual. Handwriting samples of 1500 individuals, representative of the U.S. population with respect to gender, age, ethnic groups, etc., were obtained. Analyzing differences in handwriting was done by using computer algorithms for extracting features from scanned images of handwriting. Attributes characteristic of the handwriting were obtained, e.g., line separation, slant, character shapes, etc. These attributes, which are a subset of attributes used by forensic document examiners (FDEs), were used to quantitatively establish individuality by using machine learning approaches. Using global attributes of handwriting and very few characters in the writing, the ability to determine the writer with a high degree of confidence was established. The work is a step towards providing scientific support for admitting handwriting evidence in court. The mathematical approach and the resulting software also have the promise of aiding the FDE.

Keywords:
document analysis, feature extraction, forensic science, handwriting identification, handwriting individuality

Paper ID: JFS2001227_474

regards
Chris Lowe

[Fisherman]

Ben writes:

"it irritates me to the point of desiring a vicious public boxing match that you should accuse her of such an oversight"

Hmmm, Ben - a few days ago David told me that I was lucky not to be standing opposite him, and now you are speaking about a boxing match...?

What is this all about? Are you threatening to beat me up, or what? Not very many people fancy their chances, Ben, since I am 193 centimetres tall and weigh around a 110 kilograms. Moreover, I have a boxing background.

Still, I think you would be in with a fair chance to do at least some damage, since I am also a pacifist and have been so for decades - I would not strike back if you decided to have a go. I would just restrain you.
The reason for my pacifism is that I am of the mindset that the one who deals the first blow is also the one who has realized that his arguments will do him no good, and so he resorts to less subtle methods.
...but does not all of this border on being very silly? Is this not a discussion board? I suggest we make do with that and leave the testosterone for other purposes.

"I listened to Leander before his views were distorted."

But they never were, Ben - we even have him telling us that "cannot be ruled out" is a verdict that traditionally belongs to the positive end of the scale! And that was something you categorically ruled out; to you, "cannot be ruled out" could NEVER classify a hit on the positive side. We now KNOW that this was not so, it is common procedure to classify the lowest, most careful hit on the positive side as EXACTLY "cannot be ruled out". Or is this something I have invented myself? Or something I distorted? Or did I hypnotize Leander into telling us this?
Of course not. He himself has the hit down as the lowest one on the positive side. That means that he thinks that Toppy was the witness. He is in no way sure, and apparently the match could be better than "the lowest, most careful one". And when Leander tells us that he would be surprised to have another outcome, we must realize that he would probably not be extremely surprised - but surprised nevertheless.
It is a match in Leanders wiew, it is by no means the best one he has seen, but it is good enough for him to expect it to hold water.

Fisherman

[truebluedub]

A statistical model for writer verification
Srihari, S.N. Beal, M.J. Bandi, K. Shah, V. Krishnamurthy, P.
Dept. of Comput. Sci. & Eng., Buffalo Univ., NY, USA;

This paper appears in: Document Analysis and Recognition, 2005. Proceedings. Eighth International Conference on
Publication Date: 29 Aug.-1 Sept. 2005
On page(s): 1105- 1109 Vol. 2
ISSN: 1520-5263
ISBN: 0-7695-2420-6
INSPEC Accession Number: 8732782
Digital Object Identifier: 10.1109/ICDAR.2005.33
Current Version Published: 2006-01-16
Abstract
A statistical model for determining whether a pair of documents, a known and a questioned, were written by the same individual is proposed. The model has the following four components: (i) discriminating elements, e.g., global features and characters, are extracted from each document; (ii) differences between corresponding elements from each document are computed; (iii) using conditional probability estimates of each difference, the log-likelihood ratio (LLR) is computed for the hypotheses that the documents were written by the same or different writers; the conditional probability estimates themselves are determined from labeled samples using either Gaussian or gamma estimates for the differences assuming their statistical independence; and (iv) distributions of the LLRs for same and different writer LLRs are analyzed to calibrate the strength of evidence into a standard nine-point scale used by questioned document examiners. The model is illustrated with experimental results for a specific set of discriminating elements.

kind regards
Chris Lowe

[truebluedub]

Automatic signature verification and writer identification — the state of the art
Purchase the full-text article



References and further reading may be available for this article. To view references and further reading you must purchase this article.

Réjean Plamondona and Guy Loretteb

a Laboratoire Scribens, Département de Génie Electrique, Ecole Polytechnique de Montréal, C.P. 6079, Succ. “A”, Montréal, Canada, H3C 3A7

b IRISA, Campus de Beaulieu, 35042, Rennes Cedex, France

Received 10 August 1987;
revised 22 March 1988.
Available online 19 May 2003.

Abstract

This paper presents a survey of the literature on automatic signature verification and writer identification by computer, and an overview of achievements in static and dynamic approaches to solving these problems, with a special focus on preprocessing techniques, feature extraction methods, comparison processes and performance evaluation. In addition, for each type of approache special attention is given to requirement analysis, human factors, practical application environments, and appropriate definitions and terminology. Throughout the paper, new research directions are suggested.

Author Keywords: Signature verification; Writer identification; Handwriting
Pattern Recognition
Volume 22, Issue 2, 1989, Pages 107-131

kind regards,
Chris Lowe

[truebluedub]

JSF Volume 47, Issue 5 (September 2002)




ISSN: 0022-1198
Published Online: 1 September 2002
Page Count: 8


Click here to download this paper now for $25

View License Agreement

Forensic handwriting examiners' expertise for signature comparison
Sita, J
Forensic Expertise Profiling Laboratory, School of Human Biosciences, LaTrobe University, Victoria, 3086, Australia.

Found, B
Forensic Expertise Profiling Laboratory, School of Human Biosciences, LaTrobe University, Victoria, 3086, Australia.

Found, B
Document Examination Team, Victoria Forensic Science Centre, ForensicDrive, Macleod, Victoria, 3085, Australia.

Rogers, DK
Forensic Expertise Profiling Laboratory, School of Human Biosciences, LaTrobe University, Victoria, 3086, Australia.


Abstract
This paper reports on the performance of forensic document examiners (FDEs) in a signature comparison task that was designed to address the issue of expertise. The opinions of FDEs regarding 150 genuine and simulated questioned signatures were compared with a control group of non-examiners' opinions.
On the question of expertise, results showed that FDEs were statistically better than the control group at accurately determining the genuineness or non-genuineness of questioned signatures. The FDE group made errors (by calling a genuine signature simulated or by calling a simulated signature genuine) in 3.4% of their opinions while 19.3% of the control group's opinions were erroneous. The FDE group gave significantly more inconclusive opinions than the control group.
Analysis of FDEs' responses showed that more correct opinions were expressed regarding simulated signatures and more inconclusive opinions were made on genuine signatures. Further, when the complexity of a signature was taken into account, FDEs made more correct opinions on high complexity signatures than on signatures of lower complexity. There was a wide range of skill amongst FDEs and no significant relationship was found between the number of years FDEs had been practicing and their correct, inconclusive and error rates.

Keywords:
expertise, forensic science, handwriting, questioned document examination, signature, validation

kind regards
Chris Lowe

[Archaic]

RE: Post #2012 FBI STUDY: Interesting that The US Dept of Justice sponsored the Document Examiner Study to see if expertly-trained professional "FBI Document Examiners" were significantly better at examining documents than "College-Educated Non-Experts"! (Does that mean their FBI funding was renewed for another year?)

My goodness, why didn't they do the obvious, and test themselves against "College-Educated Non-FBI Document Experts"? Where's the scientific rigor?

[truebluedub]

Progress on the problem of signature verification has advanced more rapidly in online applications than offline applications, in part because information which can easily be recorded in online environments, such as pen position and velocity, is lost in static offline data. In offline applications, valuable information which can be used to discriminate between genuine and forged signatures is embedded at the stroke level. We present an approach to segmenting strokes into stylistically meaningful segments and establish a local correspondence between a questioned signature and a reference signature to enable the analysis and comparison of stroke features. Questioned signatures which do not conform to the reference signature are identified as random forgeries. Most simple forgeries can also be identified, as they do not conform to the reference signature's invariant properties such as connections between letters. Since we have access to both local and global information, our approach also shows promise for extension to the identification of skilled forgeries.
Index Terms:
handwriting recognition; local correspondence; random forgery detection; signature verification; online applications; offline applications; forged signatures; stroke level; stroke segmentation; stylistically meaningful segments; questioned signature; reference signature; stroke features; random forgeries; invariant properties; skilled forgeries
Citation:
J.K. Guo, D. Doermann, A Rosenfeld, "Local correspondence for detecting random forgeries," icdar, pp.319, Fourth International Conference Document Analysis and Recognition (ICDAR'97), 1997

kind regards
chris lowe

[truebluedub]

And here's one on using photocopies: http://www.ajarnpat.com/data/seminar_Ex02-Eng.pdf
regards
Chris Lowe

[Fisherman]

Ben, again:

"Ah, but he never said that.
He never once claimed that Toppy was "probably" the Dorset Street witness".

He never phrased it like that, no. What he DID say though, is that it would surprise him if the signatures were not a match. And that I allow myself to clarify by saying that what he actually said at that stage was that he would be surprised if Toppy was not the man who signed the police report back in 1888.

Do you think, Ben, do you really, really think that this material does not allow me to say that Leander is of the mind that Toppy was probably the Dorset Street witness? We know that he would be surprised if that was NOT the case. Donīt you think that that being surprised if that was not the case kind of equals being of the mind that it was probable that it WAS the case?

You are free to deny this, of course. But I would regard such a denial as intellectually corrupt for obvious reasons.

You cannot just deny things that are pretty damn obvious, Ben. If a person says that one out of three objects, A, B and C, are red, and adds that A and B are NOT red, then C MUST be the red object. Admittedly, the person in question has never stated "object C is red" - but it goes without saying, Ben. Besides, when it comes to Iremonger, you seem very allowing when it comes to how much we can allow ourselves to read in, with no proof and no recorded wording AT ALL. You try to substantiate it by telling us that Martin Fido found the lady very nice and reassuring, but what kind of proof is that? You will accept his second-hand judgement on this although he is in no way an expert on handwriting, whereas you call Leander totally unreliable...?

I know Martin Fido slightly. We have a mutual friend, and Martin and I have been drinking beer and strolling through the East end together, discussing the case. He is a very amiable man and a very knowledgeable one, and I liked him thoroughly.
But that does not make him of much use to substantiate what Iremonger saw, does it? The only thing we have left of Iremonger is that it seems quite clear that she, on balance, believed that the signatures she looked at were not written by the same man. But that is ALL we have! We have no confirmation of what she saw, only your assertion that she MUST have seen the correct signatures - which, of course she "must" not have. We have no wordings, and no pointing out of any element involved in her judgement. Therefore, the only thing we can do for you in the Iremonger regard, is to gracefully accept that she reached a negative conclusion on the match she investigated. But that conclusion is of extremely poor value until we have some substantiation - that goes beyond your own good will - telling us that she saw the right material. And after that, we must also be informed what it was that made her opt for her decision.
Until that point, Iremongers verdict must be regarded with heaps of scepticism and barrels of salt - not because we in any way can allow ourselves to judge Iremonger herself, but because we MUST judge the utter incompleteness of the material involved as incredibly damning.

You have, on this thread, said that I do my job as a journalist in a bad way. But this, Ben, is EXACTLY how a journalist works - we do not buy ANYTHING until we have the sources in place and the substantiation at hand. Yo accept Iremonger as it stands, without questioning things at all, would be to dig a journalistic grave.

Fisherman

[Crystal]

Originals vs Copies

Right then

I have already said -

A copy is a copy

An original is an original

They are not comparable. One is derived from the other.

A document is not 2 dimensional. It is a 3 dimensional object. That applies whether it is an original, or a copy.

Nevertheless, a copy, however good, 'flattens' the images contained therein, obscuring detail which might otherwise be telling. A copy is not the same document, therefore, signs of its use, and of wear will be absent.

I will not know, for example, how a document has been kept by looking at a copy - at least, not nearly to the same extent. I will not know if those marks on it are ink, blood, or some other residue. I will not be able to see any evidence of pressure - from leaning on the document to write. And so on.

An example - make of it what you will;

I recently examined a foundation charter of Queen Anne. Huge thing, it was, bigger than me. It, apart from the text and the Royal Seal at the bottom, had a highly decorated border, floriated, containing little blokes and birds and what have you - all very fashionable for 1705.

Now then, I know that prior to fairly recent times - in some cases up to the 1950's, most official documents were handwritten - and in a case such as this - hand-drawn. I know this from experience of looking at such documents dating from the 7th Century onward. Pretty much always, up to the 19th Century at least, the construction lines for the placing of the text in a document, and for any decoration, can be seen, sometimes with the naked eye, or if not, under magnification.

The examination of this document, however, revealed no such construction lines. What was the answer? It had to be, rather surprisingly, that this was a printed document - an 18th Century pro-forma for officialdom. In other words, it is in itself, a copy - in this case, of a lost prototype.

Now, if I had been looking at a copy of the original charter, I still wouldn't have seen construction lines, but I couldn't have told that it was significant that I didn't see them, because they are often faint, and could easily have been lost in translation - as had actually happened between the original prototype and the 18th Century 'Copy', which then became and original Charter.

I wouldn't have known that those construction lines were never there on the Charter to begin with - I wouldn't have reached the conclusions I did.


Information is lost in copy. More information can be obtained from the original. That is why the original is better. It stands to reason. It is utterly logical. It requires no further explanation.

[Crystal]

ARE they, though, Ben? And, if they are, did Sue Iremonger actually use them?

It really is not good enough assuming that she did. Not good enough at all

Yes, they ARE. And barring flood, fire, or possibly Alien Abduction, they are likely to remain so for some time.

It may suit to believe that Iremonger did not look at the originals - as you wish, but neither has Leander, so that being the case, nobody has, have they?

So therefore, Leander's view is no better than Iremongers and we can all go home.

Agreed? No?