A Detection and Verification Based Approach for On-line Recognition of Handwritten Chinese Words


Grant Data
Project Title
A Detection and Verification Based Approach for On-line Recognition of Handwritten Chinese Words
Principal Investigator
Dr Huo, Qiang   (Principal investigator)
Duration
24
Start Date
2005-09-01
Completion Date
2007-08-31
Amount
386274
Conference Title
Presentation Title
Keywords
handwriiten Chinese word
Discipline
Others - Computing Science and Information Technology,Artificial Intelligence (Obsolete)
Sponsor
RGC General Research Fund (GRF)
HKU Project Code
HKU 7136/05E
Grant Type
General Research Fund (GRF)
Funding Year
2005/2006
Status
Completed
Objectives
(1) to investigate a new "detection and verification based approach" for on-line recognition of handwritten Chinese words; (2) to investigate new techniques for feature extraction and character modeling to improve the character classification accuracy as well as the reliability of character verification; (3) to develop a prototype system on Tablet PC, as a test-bed of new ideas, by using the most promising solution identified through the study in this project. The character vocabulary of our recognizer will include 6763 simplified Chinese characters in GB2312-80 standard, 12 frequently used GBK Chinese characters, 62 alphanumeric characters, 140 punctuation marks and symbols. We are aiming at a writer-independent character recognition accuracy of above 90% for naturally written Chinese words derived from our handwriting corpus.