A Hybrid Method for Historical Degraded Document Image
Naouel Ouafek, Mohamed-KhireddineKholladi · 2018
This paper addresses the problem of document image binarization, in which we present a new binarization method based on the use of a supervised learning support vector machine SVM, that accurately segment the text in the degraded document image. First we start by enhance the contrast of the original image. Then, a feature selection is making by using an inpainting algorithm. The last phase is the use of support vector machine to classify the pixels of the original image into "1" or "0" to construct the binary image. The suggested method has been tested and evaluated over the benchmarking in the document binarization: DIBCO'2009.