Abstract
Recently, a lot of medical documents have been computerized. On the other hand, a large amount of paper-based documents are still archived and they are not still reused effectively. Previously we discussed document image recognition methods for medical/clinical documents and their applications. We also discussed the recognition method for schema images in the document. However, annotations written in the schema made the feature vector change drastically, as a result the recognition accuracy was reduced. In this paper, we proposed a schema recognition method considering annotations. 30 kinds of actual schema images used in hospital were employed as experimental materials. After evaluation experiments, the recognition accuracy was improved to 95.27%.
