A pattern-based recognition method in online handwriting Uyghur character recognition (グラフィクスとCAD)

概要

論文の詳細を見る
This paper discusses pattern-based online handwriting Uyghur character recognition. A statistical analysis result showed that the sub-word is the basic unit of Uyghur writing. As a first step of the implementation of a sub-word-based recognition system, the one-character sub-words were recognized using Approximate Stroke Sequence String Matching method, and got a recognition rate about 97%. Experiment result proved this method also effective to the recognition of other sub-words. To get a clear idea about sub-word patterns, the sample of containing 4 million characters was collected and analyzed. Based on the analysis result, a pattern-based sub-word recognition approach is suggested. The recognition result of one-character sub-words is indicated. We discuss that it is possible to recognize 726 kinds of two-character sub-words by the recognizing half of their patterns only. In addition, it is believed that an ideal recognition rate can be obtaining by combining other features such as sub-word frequency. The recognition of other sub-words is the targets of our future work.
社団法人情報処理学会の論文
2008-08-15