Publication

A comparison of clustering methods for writer identification and verification

Bulacu, M. L. & Schomaker, L. R. B., 2005, Proceedings of the 8th International Conference on Document Analysis and Recognition (ICDAR 2005). Piscataway: IEEE (The Institute of Electrical and Electronics Engineers), Vol. II. p. 1275-1279 5 p.

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

APA

Bulacu, M. L., & Schomaker, L. R. B. (2005). A comparison of clustering methods for writer identification and verification. In Proceedings of the 8th International Conference on Document Analysis and Recognition (ICDAR 2005) (Vol. II, pp. 1275-1279). Piscataway: IEEE (The Institute of Electrical and Electronics Engineers).

Author

Bulacu, M.L. ; Schomaker, L.R.B. / A comparison of clustering methods for writer identification and verification. Proceedings of the 8th International Conference on Document Analysis and Recognition (ICDAR 2005). Vol. II Piscataway : IEEE (The Institute of Electrical and Electronics Engineers), 2005. pp. 1275-1279

Harvard

Bulacu, ML & Schomaker, LRB 2005, A comparison of clustering methods for writer identification and verification. in Proceedings of the 8th International Conference on Document Analysis and Recognition (ICDAR 2005). vol. II, IEEE (The Institute of Electrical and Electronics Engineers), Piscataway, pp. 1275-1279, 8th International Conference on Document Analysis and Recognition (ICDAR 2005), 29/08/2005.

Standard

A comparison of clustering methods for writer identification and verification. / Bulacu, M.L.; Schomaker, L.R.B.

Proceedings of the 8th International Conference on Document Analysis and Recognition (ICDAR 2005). Vol. II Piscataway : IEEE (The Institute of Electrical and Electronics Engineers), 2005. p. 1275-1279.

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

Vancouver

Bulacu ML, Schomaker LRB. A comparison of clustering methods for writer identification and verification. In Proceedings of the 8th International Conference on Document Analysis and Recognition (ICDAR 2005). Vol. II. Piscataway: IEEE (The Institute of Electrical and Electronics Engineers). 2005. p. 1275-1279


BibTeX

@inproceedings{c237408ef06c467ea6dfce9726acb9a6,
title = "A comparison of clustering methods for writer identification and verification",
abstract = "An effective method for writer identification and verification is based on assuming that each writer acts as a stochastic generator of ink-trace fragments, or graphemes. The probability distribution of these simple shapes in a given handwriting sample is characteristic for the writer and is computed using a common codebook of graphemes obtained by clustering. In previous studies we used contours to encode the graphemes, in the current paper we explore a complementary shape representation using normalized bilmaps. The most important aim of the current work is to compare three different clustering methods for generating the grapheme codebook: k-means, Kohonen SOM 1D and 2D. Large scale computational experiments show that the proposed method is robust to the underlying shape representation used (whether contours or normalized bitmaps), to the size of codebook used (stable performance for sizes from 10(2) to 2.5 x 10(3)) and to the clustering method used to generate the codebook (essentially the same performance was obtained for all three clustering methods).",
author = "M.L. Bulacu and L.R.B. Schomaker",
note = "intern. conference series Event: ICDAR 2005, Seoul, Korea",
year = "2005",
language = "English",
isbn = "0-7695-2420-6",
volume = "II",
pages = "1275--1279",
booktitle = "Proceedings of the 8th International Conference on Document Analysis and Recognition (ICDAR 2005)",
publisher = "IEEE (The Institute of Electrical and Electronics Engineers)",

}

RIS

TY - GEN

T1 - A comparison of clustering methods for writer identification and verification

AU - Bulacu, M.L.

AU - Schomaker, L.R.B.

N1 - intern. conference series Event: ICDAR 2005, Seoul, Korea

PY - 2005

Y1 - 2005

N2 - An effective method for writer identification and verification is based on assuming that each writer acts as a stochastic generator of ink-trace fragments, or graphemes. The probability distribution of these simple shapes in a given handwriting sample is characteristic for the writer and is computed using a common codebook of graphemes obtained by clustering. In previous studies we used contours to encode the graphemes, in the current paper we explore a complementary shape representation using normalized bilmaps. The most important aim of the current work is to compare three different clustering methods for generating the grapheme codebook: k-means, Kohonen SOM 1D and 2D. Large scale computational experiments show that the proposed method is robust to the underlying shape representation used (whether contours or normalized bitmaps), to the size of codebook used (stable performance for sizes from 10(2) to 2.5 x 10(3)) and to the clustering method used to generate the codebook (essentially the same performance was obtained for all three clustering methods).

AB - An effective method for writer identification and verification is based on assuming that each writer acts as a stochastic generator of ink-trace fragments, or graphemes. The probability distribution of these simple shapes in a given handwriting sample is characteristic for the writer and is computed using a common codebook of graphemes obtained by clustering. In previous studies we used contours to encode the graphemes, in the current paper we explore a complementary shape representation using normalized bilmaps. The most important aim of the current work is to compare three different clustering methods for generating the grapheme codebook: k-means, Kohonen SOM 1D and 2D. Large scale computational experiments show that the proposed method is robust to the underlying shape representation used (whether contours or normalized bitmaps), to the size of codebook used (stable performance for sizes from 10(2) to 2.5 x 10(3)) and to the clustering method used to generate the codebook (essentially the same performance was obtained for all three clustering methods).

M3 - Conference contribution

SN - 0-7695-2420-6

VL - II

SP - 1275

EP - 1279

BT - Proceedings of the 8th International Conference on Document Analysis and Recognition (ICDAR 2005)

PB - IEEE (The Institute of Electrical and Electronics Engineers)

CY - Piscataway

ER -

ID: 959881