An image processing apparatus includes an extracting unit that extracts each tablespace image from each page of image data containing plural pages read by a document reading device, a generating unit that generates each table structure data of the tables from each tablespace image extracted by the extracting unit, a discrimination unit that discriminates a connection possibility between the tables based on table structure data of the tables of each page generated by the generating unit, a determination unit that determines a connection sequence for restoring an original table by connecting each of the tables based on the connection possibility between the tables discriminated by the discrimination unit, and a restoring unit that restores data on a single table before division by connecting each of the tables based on the connection sequence determined by the determination unit.