A New Rule to Reconfirm Potential Segmentation Points with Vertexes Points of VDS
Jabril Ramdan, Khairuddin Omar, Mohammad Faidzul · Middle-East Journal of Scientific Research · 2016
In this paper present, a different rule for reconfirming the potential segmentation points of Arabic words using peaks and vertexes points of Voronoi diagrams on the baseline. The basic idea of this method is based on peaks detection and baseline. Inspired by three characteristics, we develop three steps to word and subword segmentation approach: in the word or sub-words segmentation step, a peaks detection function is adopted to model the maximum and minimum peaks. To get the model parameters, a stroke operator is utilized to extract of potential segmentation points; then a determine baseline process is developed to estimate the parameters depend on the mostly minimum peaks and determine nearest vertexes point to minimum peak on the baseline to confirm the minimum peak as segmentation point or not. Where the experimental of this method has achieved promising results with a high accuracy for Arabic characters by using four types of handwritten datasets: IFN-ENIT, AHDB, Al Hamad DB and AHDB-FTR. Early results show that the application of EDMS to MLP_ANN provides more useful results than GLCM and MOMENT in other categories, yielding results of EDMS features on MNN classifier is 95.09% for the IFN-ENIT dataset. Copyright © 2016 Praise Worthy Prize S.r.l. - All rights reserved.