A METHOD OF IMPROVING THE RATIO OF OPTICAL CONTENT RECOGNITION IN FORM-BASED PHYSICAL DOCUMENT DIGITALIZATION USING A MULTI-SPECTRAL ANALYSIS APPROACH

摘要

The proposed device (Multispectral Scanner) is used to divide the scanned document into layers. Each of these corresponds to a different medium, eg. Paper, ink printout, laser printout, dirt etc. The device utilizes the properties of light waves such as reflectance, absorption, etc. This makes it possible to identify the pigments present in the output. Available now on the market document scanners are not able to automatically separate the content for OCR (content suitable for optical character recognition) from noise and dirt. It is worth noting that it is important not only to use the proposed construction of a scanner but also the proposed algorithms which are analyzing the recorded data.