Personal identification based on handwriting
Introduction
Signature verification has been an active research topic for several decades in the image processing and pattern recognition community [1]. Despite continuous effort, signature verification remains a challenging issue. It provides a way of identifying the writer of a piece of handwriting in order to verify the claimed identity in security and related applications. It requires the writer to write the same fixed text. In this sense, signature verification may also be called text-dependent writer verification (which is a special case of text-dependent writer identification where more than one writer has to be considered). In practice, the requirement and the use of fixed text makes writer verification prone to forgery. Furthermore, text-dependent writer identification is inapplicable in many important practical applications, for example, the identification of the writers of archived handwritten documents, crime suspect identification in forensic sciences, etc. In these applications, the writer of a piece of handwriting is often identified by professional handwriting examiners (graphologists). Although human intervention in text-independent writer identification has been effective, it is costly and prone to fatigue.
Research into writer identification has been focused on two streams, off-line and on-line writer identification. On-line writer identification techniques are not well developed (as compared to on-line signature verification methods), and only a few papers (e.g. [2]) have been published on this subject. In comparison, off-line systems have been studied either as fully automated tools or as interactive tools. These systems are based on the use of computer image processing and pattern recognition techniques to solve the different types of problems encountered: pre-processing, feature extraction and selection, specimen comparison and performance evaluation.
This paper presents an off-line system based on computer image processing and pattern recognition techniques. There are two approaches to the off-line method, namely text-dependent and text-independent. Our work is a text-independent approach where a texture analysis technique is introduced. The text-independent approach uses feature sets whose components describe global statistical features extracted from the entire image of a text. Hence it may be called texture analysis approach.
Two general approaches have been proposed in the off-line method: Histogram descriptions and Fourier transform techniques. In the first case, the frequency distribution of different global and local properties is used [3]. Some of these properties are directly or indirectly related to specific features used in the forensic document analysis [4].
In the second case, Duverony et al. [5] have reported that the most important variation of the writers transfer function is reflected in the low-frequency band of Fourier spectrum of the handwriting images. Similarly, Kuckuck [6] has used Fourier transform techniques to process handwritten text as texture. The features extracted were either composed of sequences of spectrum mean values per bandwidth, polynomial fitting coefficients or a linear mapping of these coefficients. The method has been tested on a set of 800 handwriting samples (20 writers, 40 samples per writer). An overall classification rate of 90% for all features extracted was obtained.
This paper uses multichannel spatial filtering techniques to extract texture features from a handwritten text block. There are many filters available for use in the multichannel technique. In this paper we use Gabor filters, since they have proven to be successful in extracting features for similar applications [7], [8], [9], [10], [11]. We also use grey-scale co-occurrence matrices (GSCM) for feature extraction as a comparison. For classification two classifiers are adopted, namely the weighted Euclidean distance (WED) and the (K-NN) classifiers.
The subsequent sections describe the normalisation of the handwriting images, the extraction of writer features, the experimental results and finally the conclusions.
Section snippets
The new algorithm
The algorithm is based on texture analysis and is illustrated diagrammatically in Fig. 1. The three main stages are described in turn in the remainder of this section.
Experimental results
A number of experiments were carried out to show the effectiveness of the proposed algorithms. Forty people were selected, then divided into two groups (each group consist of 20 people). Examples of handwriting by these people are shown in Fig. 3.
For the purpose of the classification experiments 25 non-overlapping handwriting blocks were extracted for each person. Each sample was selected from an A4 page, scanned using a HP ScanJet4c in custom mode with extra heavy lighting, at a resolution of
Future work
The approach that has been adopted here is mainly text-independent. In the future text-dependent writer identification will be introduced. This will cover writer signature verification approaches. A comparison between the two approaches will then be drawn.
Currently our work is based on the extraction of global features, but further work will focus on the use of local features. An integrated system will be considered to combine both local and global features to produce more reliable
Conclusion
We have described a new approach for handwriting based personal identification. Most existing approaches assume implicitly that handwritten texts are fixed. The novel approach introduced in this paper eliminates such an assumption. The algorithm is based on the observation that the handwriting of different people is visually distinctive and a global approach based on textures analysis can be adopted. The approach is therefore text or content independent.
A number of experiments have been
About the Author—HUWIDA SAID received a B.Eng. honours degree in Electrical and Electronic Engineering from The University of Wales, Swansea in 1995. She is currently writing up for a PhD at The University of Reading and will graduate in July 1999. Her research interest include personal identification from facial features, handwritten characters and audio features. She is an associated member of the IEE and a member of the BCS.
References (18)
- et al.
Automatic signature verification and writer identification-the state of art
Pattern Recognition
(1989) Texture edge detection by modelling visual cortical channels
Pattern Recognition
(1995)- et al.
A review of recent texture segmentation and feature extraction techniques
CVGIPImage Understanding
(1993) - et al.
Automatic orientation and skew angle detection for binary document images
Pattern Recognition
(1994) - et al.
Automatic handwriting identification based on the external properties of the samples
IEEE Trans. Systems Man Cybernet
(1983)- V. Klement, K. Steine, R. Naske, The application of image processing and pattern recognition techniques to the forensic...
Handwriting synthesis and classification by means of space-variant transform and Karhunen–Loeve analysis
J. Opt. Soc. Am.
(1975)- W. Kuckuck, Writer recognition by spectra analysis, in: Proceedings of the International Conference On Security Through...
Cited by (279)
Siamese-based offline word level writer identification in a reduced subspace
2024, Engineering Applications of Artificial IntelligenceUtilization of information from CNN feature maps for offline word-level writer identification
2024, Expert Systems with ApplicationsRecent trends in mathematical expressions recognition: An LDA-based analysis
2023, Expert Systems with ApplicationsA bibliometric analysis of off-line handwritten document analysis literature (1990–2020)
2022, Pattern RecognitionWriter Identification using Deep Learning with FAST Keypoints and Harris corner detector
2021, Expert Systems with Applications
About the Author—HUWIDA SAID received a B.Eng. honours degree in Electrical and Electronic Engineering from The University of Wales, Swansea in 1995. She is currently writing up for a PhD at The University of Reading and will graduate in July 1999. Her research interest include personal identification from facial features, handwritten characters and audio features. She is an associated member of the IEE and a member of the BCS.
About the Author—TIENIU TAN received his B.Sc. (1984) in Electronic Engineering from Xi'an Jiaotong University, China, and M.Sc. (1986), DIC (1986) and Ph.D. (1989) in Electronic Engineering from Imperial College of Science, Technology and Medicine, London, England.
In October 1989, he joined the Computational Vision Group at the Department of Computer Science, The University of Reading, England, where he worked as Research Fellow, Senior Research Fellow and Lecturer. In January 1998, he returned to China to take up a professorship at the National Laboratory of Pattern Recognition located at the Institute of Automation of the Chinese Academy of Sciences, Beijing, China.
Dr. Tan has published widely on image analysis and computer vision. He is a Senior Member of the IEEE and an elected member of the Executive Committee of the British Machine Vision Association and Society for Pattern Recognition (BMVA). He is the Asia Editor of the International Journal of Image and Vision Computing and an Associated Editor of the Pattern Recognition journal. His current research interests include speech and image processing, machine and computer vision, pattern recognition, multimedia, and robotics.
About the Author—KEITH BAKER Studied both Electronics Engineering and Mathematical Physics at the undergraduate level before receiving the PhD degree in physics in 1969 from the University of Sussex, Brighton, England.
After a period as postdoctoral fellow, he worked as a Systems Analyst on real-time air traffic control systems with the Software Systems Division of the Plessey Company. Later he was with the Burroughs Corporation, Detroit, MI, as a Supervisor of Software Engineering and Project Manager.
He has also spent some time as Lecture in Computer Science at the University of Sussex before taking up an appointment as Professor of Information Engineering Systems and head of the Electrical and Electronic Engineering Department at the University of Plymouth, England. In 1986 he moved to the University of Reading as full Professor of Computer Science, later to become the head of the Computer Science Department. He is Currently Dean of the Faculty of Sciences. His research interest include contributions to software engineering, computer vision, and intelligent systems.
Dr. Baker is the Editor-in Chief of the international Journal Image and Vision Computing, a Fellow of the IEE, and a member of the IEEE, the ACM and BCS.