Recognizing broken characters in Thai Historical documents

摘要

One of the biggest challenges in restoring historical documents is to achieve a high level of OCR accuracy. The main characteristic inherent to these valuable but degraded documents is the abundant presence of broken characters. This paper represents this problem as a mathematical model. We also propose a novel solution based on set-partitions to recognize broken characters in Thai Historical documents. Experiments based on this solution have been performed and the results are very promising.