The New Text and Graphical Input Device: Compact Biometrical Data Acquisition Pen

1
The New Text and Graphical Input Device Compact
Biometrical Data Acquisition Pen

• Václav Matoušek, Ondrej Rohlík,
• Pavel Mautner, Jürgen Kempf

Department of Computer Science and
Engineering University of West Bohemia in Pilsen
2
Outline

• Data acquisition device The BiSP pen
• Handwritten text recognition
• Hidden Markov models
• Experimental results
• Future Work

3
Input Devices Overview

• off-line (static)
• scanners
• cameras
• on-line (dynamic)
• electronic pens
• digitizers, tablets
• cameras
• mouse

4
Input Device The BiSP Pen

• Electronic pen is used for data acquisition

built at the University of Applied Sciences in
Regensburg, Germany

5
Input Device Writing
6
Input Device Signals
7
Handwritten Text Recognition

• Objective To convert handwritten sentences or
  phrases in analog form (off-line or on-line
  sources) into digital form (ASCII or Unicode).
• isolated character recognition (TM, DTW, NN)
• word recognition (HMMs)
• gesture recognition

8
Handwritten Text
hand printed characters spaced discrete
characters cursive script words
9
Input Device Signals
10
Signal Description

• Pairs of x and y signals are transformed
  into sequence of primitives

Primitive (observation) Signal trend Signal trend
Primitive (observation) x y
1 ? ?
2 ? ?
3 ? ?
4 ? ?
11
Hidden Markov Models

• left-to-right model
• (used mostly in speech recognition)

12
Hidden Markov Models

• Training Baum-Welch algorithm
• Recognition Backward algorithm
• Matrices that describes the word model (A, B, ?)
  are decomposed after the training one model for
  each letter is obtained

13
Word HMM Decomposition
14
Word HMM Composition
15
Experimental Results

• method have been tested on three independent data
  sets of various sizes
• limited number of letters used in our data sets
  15
• reduced complexity of tagging the training set

Vocabulary size 1649 2198 5129
Recognition rate () 88 90 82
Recognition time (min) 17-26 27-49 360
16
Future Work

• to speed up the algorithm to achieve
  real-time recognition
• incorporation of language models to improve the
  recognition rate
• special attention will be paid to signature
  analysis and signature verification

17
(No Transcript)