Form classification using DP matching

Abstract

In order to process many kinds of form document, form classification must be performed in the first place. However, a substantial amount of time is required to analyze and recognize such form documents according to the increase of the number of registered models. To solve this problem, we suggest document understanding with a partial matching method in which form structure recognition and form classification are performed for only some areas of the input form. In this case, the definition of the matching areas and form classification are performed by using a DP matching algorithm. By using this approach, the recognition of an input form and the extraction of field items can be processed at, a high recognition rate in a reasonable time span. In terms of applicability, the experimental recognition rate and processing time seem to be encouraging. © 2000 ACM.