Fan Beam Projection Based Features to Recognize Handwritten Kannada Numerals
Mamatha H.R, Srikanta Murthy K, S Sudan, V. Ganesh Raj, Sumukh S Jois · 2011
The traditional goal of the feature extractor is to characterize an object by making numerical measurements. Good features are those whose values are similar for objects belonging to the same category and distinct for objects in different categories. In this paper the Fan beam projection, a variation of Radon transform is proposed for extracting the features of handwritten Kannada numerals. Features were computed using fan-beam geometry. For Fan-beam, 55 diverging beams are taken. Fan-beam takes projections at different angles by rotating the source around the center pixel at θ degree intervals. These projection data is considered as feature vector. For Fan-beam the average of the projections of one direction was taken which is the average of 55 parallel projections. Hence size of feature vector for one numeral is 1×360. For classification K-Nearest Neighbor (K-NN) classifier is used. The proposed algorithm is experimented on nearly 1000 images of handwritten Kannada numerals and appreciable results are obtained.