Forensics: Signature Analysis
Forensic examination and analysis of writer identification and verification are reported as having "applications in various domains. Especially in the field of biometric recognition. Forensic Document Examination (FED) is a process that clarifies the writer of a handwritten document. Applications included are signatures verification. FED is such that is reported to have been based primarily on manual examination by experts with only recent automation of analysis methods being initiated.
State of the Art Systems
Two state of the art systems presently exist to "address the tasks of writer identification and verification" and include: (1) the system that uses Hidden Markov Models (HMMs); and (2) the system based on Gaussian Mixture Models (GMMs) to model a person's handwriting. (Sclapbach and Bunke, nd) In addition, there are reported to be "various feature selection methods applied to improve the performance of an existing off-line writer identification system. The performance of the system is significantly impro0vec by selected a good subset of the original feature set. The methods discussed can be potentially applied to any set of features and are thus not restricted to off-line handwriting." (Sclapbach and Bunke, nd)
III. Online Signature Analysis
On-line handwriting data requires the use of a GMM-based system for identification of the writer. It is reported that as a biometric, signature 'Has a number of properties that make it especially appealing. Signature is a behavioral biometric where the users is asked to perform a specific action. Therefore, it is impossible to be stolen and hard to be copied. A signature can be easily, inexpensively, and non-invasively captured. Especially as mobile devices such as smart phones, PDAs and Tablet PC's with built...
In a report on the performance of forensic document examiners in a signature comparison task that was designed to address the issue of expertise stated is "150 genuine and simulated questioned signatures were compared with a control group of non-examiner's opinions, and the results show that the expertise of the FEDs were statistically better than the control group at determining whether the questioned signatures were genuine. (Sita, Found, & Rogers, 2002) In another study, it is reported that proposed are "new machine learning strategies for person identification which can be used in several biometric modalities such as friction ridges, handwriting, signatures and speech. The biometric or forensic performance task answers the question of whether or not a sample belongs to a known person." (Srinivasan, Beal, & Srihari, 2005)
IV. Proficiency Skills of Forensic Experts
It is reported that a test of proficiency in the skill of Handwriting Identification has been developed by the Forensic Expertise Profiling Laboratory at La Trobe University in Victoria, Australia. The testing program involves a two-part test. It is reported that for 2004, "in part one of the test 100 prepared questioned signatures had to be compared to a group of known signature specimens. In part two, given later in the year, questioned handwritten text had to be evaluated in comparison to known handwritten text. This would be similar to an anonymous note problem in casework. An additional test addressed questions of Line Sequence (which intersecting line was place above the other)." (Will, 2008) The report states that the known and questioned material was prepared by subjects drawn from "graduate students, staff members of…
