119 research outputs found
Support vector machines for the classification of western handwritten capitals
No full textItem does not contain fulltext620 p
Allograph based writer identification, handwriting analysis and character recognition
Full text linkContains fulltext :
82000.pdf (Publisher’s version ) (Open Access)In this thesis, techniques and features are described that were developed for the automatic comparison of handwritten characters by first matching them to prototypical character shapes (allographs). These techniques and features were evaluated in experiments simulating different real-world applications. The majority of the experiments regard forensic writer identification, where the objective is to find the writer of a piece of handwriting by comparing it to a large set of handwritten documents of which the writer is already known. The assumption is that if two documents contain many similar allographs, they may have been produced by the same writer. In the experiment described, it is demonstrated that using the techniques and features, it is indeed possible to match the correct writer with a piece of unknown handwriting. Other experiments were performed to evaluate the usefulness of the techniques and features for the classification of hand-drawn symbols and characters in differentive is not to find out who produced the writing, but what it represents) and the analysis of children's handwriting to diagnose Developmental Coordination Disorder.Radboud Universiteit Nijmegen, 04 oktober 2010Promotores : Schomaker, L.R.B., Desain, P.W.M. Co-promotor : Vuurpijl, L.G.227 p
Spiking neuron-inhibitory Interneuron (NiN) oscillator ensemble model for modeling (handwritten) time series
No full textBackup (.tgz) of directory with software and data for the modeling of a spike-train ensemble based on neuron-inhibitory interneuron (NiN) pairs. The test case was to model handwritten pen-tip trajectories. This could be achieved with a reasonable error in 500 epochs, where Jordan/Elman MLPs (RNN) needed hundreds of thousands of epochs. The work was later published as an article:
Schomaker, L.R.B. (1992). A neural-oscillator model of temporal pattern generation. Human Movement Science, 11, 181-192.</p
Analysis of texture and connected-component contours for the automatic identification of writers
Full text linkRecent advances in 'off-line' writer identification allow for new applications in handwritten text retrieval from archives of scanned historical documents. This paper describes new algorithms for forensic or historical writer identification, using the contours of fragmented connected-components in free-style handwriting. The writer is considered to be characterized by a stochastic pattern generator, producing a family of character fragments (fraglets). Using a codebook of such fraglets from an independent training set, the probability distribution of fraglet contours was computed for an independent test set. Results revealed a high sensitivity of the fraglet histogram in identifying individual writers on the basis of a paragraph of text. Large-scale experiments on the optimal size of Kohonen maps of fraglet contours were performed, showing usable classification rates within a non-critical range of Kohonen map dimensions. The proposed automatic approach bridges the gap between image-statistics approaches and purely knowledge-based manual character-based methods. (c) 2006 Elsevier B.V. All rights reserved
Adaptive Recognition of Online, Cursive Handwriting
No full textIn earlier studies, a stroke-oriented recognizer (VHS) of on-line cursive handwriting is reported [Thomassen et al., 1988; Schomaker & Teulings, 1990; Teulings et al., 1990; Schomaker & Teulings, 1992; Schomaker, 1993]. This system uses a movement-theoretical segmentation into strokes as the starting point of the recognition process. The pen-tip trajectory of a written word is low-pass filtered, and geometrically normalized with respect to size and slant. The absolute velocity of the pen-tip displacement is calculated, and the signal is segmented in strokes, each stroke being the trajectory between two robust minima in the absolute velocity [Teulings et al., 1987]. Strokes are characterized by feature vectors that are clustered using a Kohonen Self-Organizing Map as a feature quantizer. In the current system, as opposed to earlier versions, a number of typical problems in connected-cursive and mixed-cursive script recognition are dealt with, such as t-bar crossing,..
Advances in Writer Identification and Verification
No full textThe behavioral-biometrics methods of writer identification and verification are currently enjoying renewed interest, with very promising results. This paper presents a general background and basis for handwriting biometrics. A range of current methods and applications is given. Results on a number of methods are summarized and a more in-depth example of two combined approaches is presented. By combining textural, allographic and placement features, modern systems are starting to display useful performance levels. However, user acceptance will be largely determined by explainability of system results and the integration of system decisions within a (Bayesian) framework of reasoning that is currently becoming forensic practice
Reading Systems: An introduction to Digital Document Processing
No full textAbstract. As an introduction to the area of digital document processing we first take a few steps back and take a look at the purpose of digital document processing. Subsequently a detailed comparison between the human and the artificial reading system is made. Finally, the chapter provides an overview on the book as a whole. Methods for the creation and persistent storage of text [10] have existed since the Mesopotamian clay tablets, the Chinese writings on bamboo and silk as well as the Egyptian writings on papyrus. For search and retrieval, methods for systematic archiving of complete documents in a library were developed by monks and by the clerks of emperors and kings in several cultures. However, the technology of editing an existing document by local addition and correction of text elements has a much younger history. Traditional copying and improvement of text was a painstakingly slow process, sometimes involving many man years for one single document of importance. The invention of the pencil and eraser in 1858 was one of the signs of things to come. The advent of the typing machine by Sholes in 1860 allowed for faster copying and a simultaneous on-the-fly editing of text. The computer, finally, allowed for a very convenient processing of text in digital form. However, even today, methods for generating a new document are still more advanced and mature than are the methods for processing an existing document
GIWIS Writer Identification Tool - Distribution for Windows - including setup.exe and example data fromFiremaker P1
No full text<p>Giwis - Snelkoppeling.lnk<br>
License-documentation-GIWIS-v3.1c-beta.pdf<br>
enfhex-GIWIS-2011b.pdf<br>
experiment/<br>
experiment.gds<br>
firemaker/ (Note: this is not the complete Firemaker data set, just P1)<br>
firemaker.gds<br>
setup.exe</p>
<p> </p>GIWIS v3.1 - beta
Groningen Intelligent Writer Identification System
Documentation
V3.1c-draft
Lambert Schomaker
November 2011
September 2012
The GIWIS program is an exploratory software tool for non-commercial applications in a forensic or paleographic context. No warranties can be given concerning reliability of matching results for handwritten documents. The user is responsible for the collection of statistical reference material for calibration of GIWIS over several years of usage, using his/her own reference collection of handwritten image samples, consisting of minimally several hundreds, preferably thousands of images of extracted, pure-handwriting samples of sufficient and standardized quality.
This GIWIS release (3.1) is intended for 'beta' testing by selected knowledgeable workers in forensic of paleographic science.
Conditions of Use:
User will:
Produce a scientific citation to the following article when reporting on the use of GIWIS:
A.A. Brink, J. Smit, M.L. Bulacu, L.R.B. Schomaker, Writer identification using directional ink-trace width measurements, Pattern Recognition, In Press, Corrected Proof, Available online 18 July 2011,
ISSN 0031-3203, DOI: 10.1016/j.patcog.2011.07.005.
(http://www.sciencedirect.com/science/article/pii/S0031320311002810
- …
