IMPROVING THE IMAGE BINARIZATION PROCESS BY USING A WEIGHTED-VOTING SCHEME
Cristian-Daniel Pobleanu, Andrei Dumitrescu, Ștefania Cristea, Giorgiana Violeta Vlăsceanu · 2020
An essential step in the OCR (Optical Character Recognition) flow is the stage of image binarization. There are considerable algorithms used for the stage of binarization. The significant difference between them is the method that computes the pixel threshold value. These algorithms can categorize into two groups: those that require a single threshold value for the whole image (which can recognize larger areas or objects) and those that use a threshold obtained from the pixel's local space (pixels in the neighborhood of the given pixel and that can identify smaller clusters of pixels). The objective of this paper is to present a voting-based image binarization algorithm that makes use of results from both categories of algorithms previously mentioned.