Data Collection and Image Processing System for Ancient Arabic Manuscripts
This paper presents a general-purpose data collection system that combines a DSLR camera with directional LED lamps in order to capture a large quantity of high-resolution manuscript images in such a way as to maximize the speed of data collection while minimizing time and the need for specialized equipment. By integrating custom image processing software, the captured document images are mapped to lie on a planar surface, thereby enabling the application of more sophisticated computer vision algorithms. For this purpose, we also introduce an optional binarization tool that allows researchers to perform basic image pre-processing to simplify later analysis. The hardware setup and software tools presented in this paper can be combined to yield a simple system capable of producing large image datasets for use in document analysis research projects.
- Computer Science & Engineering [598 items ]