print

Evaluation of classification algorithms for automatic language transcription system for ancient handwriting Javanese manuscripts

---

As the Ancient Javanese Manuscript grows old, the risk of the document disappearingrnbecomes greater and greater. Digitalization of these invaluable documents isrninevitable. Once the documents are digitalized, they could be presented to furtherrnimage and pattern processing, such as Optical Character Recognition (OCR). As partrnof this project, this study is focused to evaluate the performance of severalrnclassification algorithms on handwriting Javanese characters. The input informationrnof the classifiers was MESH and LLD features of the character image. The algorithmsrnused in the experiments included k-Nearest Neighbor, Artificial Neural Networkrn(ANN) and Support Vector Machine (SVM), which serve as the classification part ofrnthe OCR. Comparing the three algorithms is expected to clarify the characteristics ofrnthose algorithms, and selecting the appropriate one for OCR implementation. We alsornevaluated a divide & conquer model by combining clustering and classificationrnalgorithm. The aim of the proposed approach is to make the model suitable for largernscale handwriting database by roughly partitioning the vector space of the data,rnfollowing by a finer classification modules.

---

Availability

B00665

(wh)

Available

Detail Information

Series Title

-

Call Number

665

Publisher

: Swiss German University., 2010

Collation

-

Language

English

ISBN/ISSN

-

Classification

NONE

Content Type

-

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

IT

Specific Detail Info

-

Statement of Responsibility

-

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment