Segmentation of Arabic Words Using Area Voronoi Diagrams andNeighbours Graph

摘要

The Voronoi neighbourhood comes from the Graphs (G), accompanied with the details concerning the correlations and neighbouring Gs lower, upper and centre points, with regards to the word group is performed. The problem entails determining the neighbours for segmenting G to Voronoi Diagrams? (VD) usage for acquisition of Voronoi Edge (VE) which with the help of neighbours sets apart components. Importantly, VE effectively segments fully neighbours however, it does not effectively segment partly neighbour. In this paper the issues addressed by the neighbourhood comprise the graph?s VD area together with the VE function as non-linear segmentation. Later, segmentation is carried out by choosing appropriate sites bordering the Voronoi. This makes the G information alongside the candidate character recognition distance to be used. Certain Arabic datasets including APTI, AHDB and IFN-ENIT applies the process of segmentation. The appropriate technique has been analysed on various Arabic font images. The experiment finally gives results that indicate right outcomes with considerable precision which is strong on varying text orientation, skew angles, types and sizes.