A framework for objective performance evaluation for digital font reconstruction

摘要

Optical Character Recognition (OCR) systems have been around for a long time. However, the need for such systems goes further than just the need to convert a scanned document into a text file. There is also a need to produce a more appealing and refine reconstructed fonts. Recently, many research works in font reconstruction have shown successful and encouraging results. However, the usual way of evaluating the performance of font reconstruction is through visual inspection which is very subjective in nature, and arduous for a more perfect and objective evaluation. Therefore, the main objective of this paper is to present a framework for an objective performance evaluation method that can be used to compare a reconstructed font with its original counterpart. Thus, it will involve studying the nature of the current and existing digital fonts and then formulate a method to compare character glyph of the created font with its corresponding original font. From this framework, it is expected that font reconstruction would be made more refine and the evaluation can be made more objective and less arduous than using the human eye, and old and rare documents can be more readily available for users in digital format with better precision in particular on tablets and other resource-constrained devices.