Handwritten Kannada Numeral Recognition based on the Curvelets and Standard Deviation
Mamatha H.R1,
Sucharitha S 2, and
Srikanta Murthy K 2
1. Department of ISE, PES Institute of Technology, Bangalore, India
2. Department of CSE, PES School of Engineering, Bangalore, India
2. Department of CSE, PES School of Engineering, Bangalore, India
Abstract—A feature extractor is generally used tocharacterize an object by making numerical measurementsof the object. Features whose values are similar for objectsbelonging to the same class and dissimilar for objects indifferent classes are considered as good features. In thispaper, an attempt is made to develop an algorithm for therecognition of handwritten Kannada numerals using fastdiscrete curvelet transform. Curvelet coefficients areobtained by applying the curvelet transform with differentscales. Standard deviation is applied to the coefficientsobtained and the result of this is used as the feature vector.A k-NN classifier is adopted for classification. The proposedalgorithm is experimented on 1000 samples of numerals.The system is seen to deliver reasonable recognitionaccuracies for different scales with 90.5% being the highestfor Scale 3.
Index Terms—Curvelets, standard deviation, handwrittenkannada numeral recognition, k-NN classifier
Cite: Mamatha H.R, Sucharitha S, and Srikanta Murthy K, "Handwritten Kannada Numeral Recognition based on the Curvelets and Standard Deviation," International Journal of Signal Processing Systems, Vol. 1, No. 1, pp. 74-78, June 2013. doi: 10.12720/ijsps.1.1.74-78
Index Terms—Curvelets, standard deviation, handwrittenkannada numeral recognition, k-NN classifier
Cite: Mamatha H.R, Sucharitha S, and Srikanta Murthy K, "Handwritten Kannada Numeral Recognition based on the Curvelets and Standard Deviation," International Journal of Signal Processing Systems, Vol. 1, No. 1, pp. 74-78, June 2013. doi: 10.12720/ijsps.1.1.74-78