Table Rows Segmentation

We consider the Document Understanding problem of segmenting tables in rows. We propose a method that first enumerates virtual row separator candidates and then select the correct ones thanks to a classification task, solved using supervised structured machine learning. Interestingly, the task is the joint-classification of virtual separators and real text lines. We describe and tested several alternative candidate generation methods and report the results of our experiment for each, on two different types of registry books from the 19th century.

I. Introduction

Over the last decades, numerous digitization campaigns have started to provide online access to a large volume of archival documents. For documents such as registry or census books, a vast majority are not reports or narratives, but tables, which can be either fully drawn and written by hand or use printed forms. Today, archival institutions are still facing major issues when trying to perform Information Extraction on these handwritten archival documents.