Cuneiform symbols recognition by support vector machine (SVM)

Cuneiform character recognition represents a complex problem in pattern recognition as result of problems that related to style of this type of writing and the diversity of its features according to distortion and shadows problems. This research proves that polygon approximation method is an optimal feature extra action method , which has been adopted for recognition task compeer with elliptic Fourier descriptor, according to the achieved high accuracy recognition results after applying multiple classes of support vector machine classifier along with depending on its discriminate functions .This work is applied by using two Data set , the first one contains 320 images of cuneiform symbols patterns for evaluate the optimal feature extraction method. The second contains 240 images of cuneiform characters to evaluate the recognition system, agents training dataset consists of 2D four triangular patterns.