Recognition experiments of cursive dynamic handwriting with self-organizing networks

doi:10.1016/0031-3203(93)90172-S

Pattern Recognition

Volume 26, Issue 3, March 1993, Pages 451-460

https://doi.org/10.1016/0031-3203(93)90172-S Get rights and content

Abstract

The aim of the paper is to assess the feasibility of using self-organizing methods for the development of a recognition system of cursive, dynamic handwriting for interactive applications with a large dictionary. A prototype system based on segmentation into motoric strokes and concurrent segmentation/classification of allographs by means of a set of allographic maps is developed. In the initial implementation, based on Kohonen's SOMs (self-organized maps), a 70% user-specific word recognition rate with a 4k-words dictionary is approached. From this, indications are derived for a modified neural recognizer (SOC: self-organized classifier) that is still based on self-organization but is more flexible (dynamic network size and topology) and can support incremental learning. The new model, together with improved pre-processing methods, could overcome the 80% mark in a pilot study with three subjects.

References (33)

R. Plamondon et al.
Automatic signature verification and writer identification: the state of the art
Pattern Recognition
(1989)
P. Morasso et al.
How a discontinuous mechanism can produce continuous patterns in trajectory formation and handwriting
Acta Psychologica
(1983)
C.J. Wells et al.
Fast dictionary look-up for contextual word recognition
Pattern Recognition
(1990)
B. Taconet et al.
A new global off-line recognition method for handwritten words
P. Morasso et al.
Trajectory formation and handwriting: a computational model
Biol. Cybern.
(1982)
C.C. Tappert et al.
The state of the art in on-line handwritten recognition
IEEE Trans. Pattern Analysis Mach. Intell.
(1990)
R. Plamondon et al.
An evaluation of motor models of handwriting
IEEE Trans. Syst. Man Cybern.
(1989)
R. Plamondon et al.
On the automatic extraction of biomechanical information from handwriting signals
IEEE Trans. Syst. Man Cybern.
(1991)
Y. LeCun et al.
Backpropagation applied to handwritten zip code recognition
Neural Computation
(1989)
G.L. Martin et al.
Recognizing handprinted letters and digits using back-propagation learning
Neural Computation
(1991)

Y. Kimura

Distorted handwritten kanji character pattern recognition by a learning algorithm minimizing output variation

M. Garris et al.

Methods for enhancing neural network handwritten character recognition

I. Guyon et al.

Design of a neural network character recognizer for a touch terminal

Pattern Recognition

(1991)

A. Waibel

Modular construction of time-delay neural networks for speech recognition

Neural Computation

(1989)

R.M. Bozinovic et al.

Off-line cursive script word recognition

IEEE Trans. Pattern Analysis Mach. Intell.

(1989)

K. Fukushima et al.

Recognition and segmentation of connected characters in cursive handwriting with selective attention

Cited by (29)

How to design finger input of Chinese characters: A literature review
2014, International Journal of Industrial Ergonomics
Citation Excerpt :
Handwriting satisfaction is affected by script style too. In psychological research, practicing Chinese calligraphic handwriting which is called as Shufa in Chinese is very beneficial for the health and emotional calmness (Eldridge et al., 1984; Kao, 2006; Morasso et al., 1993). So script style leads to user satisfaction.
A comprehensive literature review is conducted on the considerations of the human finger and Chinese character in finger interaction for Chinese input. The results of this study indicate the importance and influences of characteristics of the human finger and Chinese characters on Chinese handwriting. Moreover, fifteen design guidelines are developed based on the literature review to improve the Chinese handwriting performance (i.e. input time, accuracy, error rate, mental workload, satisfaction and physical fatigue) and subjective ratings. Seven guidelines are developed regarding properties of the human finger (i.e. finger size, contact area, motor skill, texture perception and input position). Another eight guidelines are related to characteristics of Chinese characters (i.e. direction, structure, script style, complexity and semantics). These results are in great help of improving the Chinese handwriting performance and user experience. The guidelines are going to help product designers develop a user-friendly Chinese handwriting system.
The literature review results give the theoretical support on developing a user-friendly Chinese handwriting system. Fifteen practical design guidelines based on the literature review are beneficial to improving handwriting performance and subjective ratings and consequently product design.
On-line Arabic handwriting recognition system based on visual encoding and genetic algorithm
2009, Engineering Applications of Artificial Intelligence
One of the most promising methods of interacting with small portable computing devices, such as personal digital assistants, is the use of handwriting. In order to make this communication method more natural, we propose to observe visually the writing process on an ordinary paper and to automatically recover the pen trajectory from numerical tablet sequences. On the basis of this work, we developed a handwriting recognition system based on visual coding and genetic algorithm “GA”. The system is applied on Arabic script. In this paper, we will present the different steps of the handwriting recognition system. We focus our contribution on the encoding system and the fitness function conception used as basic steps of the GA. A new approach based on visual indices similarity is developed to calculate the evaluation function. We optimize the times cooling of our system to give the final output (proposed words). Several experimentations are developed using an Arabic data set words extracted from “LMCA” database elaborated in our laboratory by 24 participants. The results obtained are very promising and prove that our new method based on hybridization between visual codes and GA is a powerful method.
On-line handwritten digit recognition based on trajectory and velocity modeling
2008, Pattern Recognition Letters
Citation Excerpt :
This method is applied in order to organize the input prototype vectors. The advantages of this method are: This algorithm is a powerful tool for giving the real classes’ number and not the semantic one (Kiviluoto, 1996; Mezghani et al., 2002; Morasso et al., 1993). In fact, it is an unsupervised algorithm.
The handwriting is one of the most familiar communication media. Pen based interface combined with automatic handwriting recognition offers a very easy and natural input method. The handwritten signal is on-line collected via a digitizing device, and it is classified as one pre-specified set of characters. The main techniques applied in our work include two fields of research. The first one consists of the modeling system of handwriting. In this area, we developed a novel method of the handwritten trajectory modeling based on elliptic and Beta representation. The second part of our work shows the implementation of a classifier consisting of the Multi-Layers Perception of Neural Networks (MLPNN) developed in a fuzzy concept. The training process of the recognition system is based on an association of the Self Organization Maps (SOM) with Fuzzy K-Nearest Neighbor Algorithms (FKNNA). To test the performance of our system we build 30,000 Arabic digits. The global recognition rate obtained by our recognition system is about 95.08%.
Automatic extraction of human-recognizable shape and execution prototypes of handwritten characters
2003, Pattern Recognition
The novel prototype extraction method presented in this paper aims to advancing in the comprehension of handwriting generation and improving on-line recognition systems. The extraction process is performed in two stages. First, using Fuzzy ARTMAP we group character instances according to classification criteria. Then, an algorithm refines these groups and computes the prototypes. Experimental results on the UNIPEN international database show that the proposed system is able to extract a low number of prototypes that are easily recognizable. In addition, the extraction method is able to condense knowledge that can be successfully used to initialize an LVQ-based recognizer, achieving an average recognition rate of 90.15%, comparable to that reached by human readers.
Recognizing Thai handwritten characters and words for human-computer interaction
2001, International Journal of Human Computer Studies
Normally, people use a keyboard to interact with a computer. This type of interaction has two main problems; typing speed and typing error. This paper proposes a non-keyboard computer interaction by using a write-pen or mouse to write Thai handwritten characters and words, using a feature-based, fuzzy logic and object-oriented approach (FBFLOOA) to recognize on-line handwritten Thai characters and words. The feature-based concept is used to extract handwritten character features, the fuzzy logic set is used to identify uncertain handwritten character shapes and the object-oriented approach is used to analyse, design and implement a handwritten character and word recognition program.
Two phases of Thai handwritten character and word recognition are proposed. The first phase uses only the FBFLOOA to recognize a handwritten character and the second phase uses FBFLOOA combined with a Thai dictionary file to seek a correct answer for a rejected recognition character. The first phase experimental results show a recognition accuracy of 89.24%, 9.20% misrecognition and 1.56% rejection. The second phase precision results are 97.82%, 0.62% misrecognition and 1.56% rejection. Both phases have an average recognition speed of 6.72s per character. The FBFLOOA-executed program size is 189 KB and the Thai dictionary file is 853 KB, which makes FBFLOOA available for notebooks, mobile phones, calculators and pocket computers.
Modelizing character allographs in omni-scriptor frame: A new non-supervised clustering algorithm
2000, Pattern Recognition Letters
The “problem of the allographs” specific of the dynamic handwriting in omni-scriptor context renders the implementation of “classical” clustering algorithms particularly delicate because it introduces the notion of heterogeneous classes characterized by strongly variable example densities. We propose here a hybrid clustering algorithm combining both a prototype placement stage and an adaptation stage. The process reduces drastically the number of references to be examined during a k-nn classification while preserving to the classifier a high level of performance. The experience has been driven on an extensive alphabet including 80 classes. Recognition rates, evaluated on nearly 35 000 examples from the UNIPEN basis show the reliability of the modelization.

View all citing articles on Scopus

View full text

Recognition experiments of cursive dynamic handwriting with self-organizing networks

Abstract

Pattern Recognition

Acta Psychologica

Pattern Recognition

A new global off-line recognition method for handwritten words

Trajectory formation and handwriting: a computational model

Biol. Cybern.

The state of the art in on-line handwritten recognition

IEEE Trans. Pattern Analysis Mach. Intell.

An evaluation of motor models of handwriting

IEEE Trans. Syst. Man Cybern.

On the automatic extraction of biomechanical information from handwriting signals

IEEE Trans. Syst. Man Cybern.

Backpropagation applied to handwritten zip code recognition

Neural Computation

Recognizing handprinted letters and digits using back-propagation learning

Neural Computation

Distorted handwritten kanji character pattern recognition by a learning algorithm minimizing output variation

Methods for enhancing neural network handwritten character recognition

Design of a neural network character recognizer for a touch terminal

Pattern Recognition

Modular construction of time-delay neural networks for speech recognition

Neural Computation

Off-line cursive script word recognition

IEEE Trans. Pattern Analysis Mach. Intell.

Recognition and segmentation of connected characters in cursive handwriting with selective attention