Regresar

Uniform vs full height maps using a time of flight Device for dewarping book spread images in the design of an automated book reader

Abstract:

This study focuses on effects of uniform and full height map correction methods for dewarping book spread images in an automated book reader design for individuals with visual impairment and blindness. The design concept could also be applied to address the challenging process of book digitization. The method is dependent on the geometry of the book reader setup for acquiring the 3-D maps that yield a high reading accuracy. The experiments were performed on a testing dataset consisting of 142 pages with their corresponding depth maps that were extracted. The accuracy of the book spread images was quantified and measured by introducing the corrected images to an Optical Character Recognition engine. Initially, the book spreads were tested by placing them with a standard alignment which yielded an average accuracy of 95.55% and 96.11% with the uniform maps and the full height maps, respectively. Rotations of the book spreads are introduced in a separate test to see if the two proposed methods are tolerant to unsuspected misaligned placements of the book. These tests yield an average accuracy of 90.63% for the corrections with a uniform map and 94.75% with the full height maps.