Research on Handwritten Numeral Character Recognition Based on Heuristic GA-SVM
YE Jian-ying · Computer Technology and Development · 2012
The technology of numeral character recognition is an important research direction in image processing area and the application in various fields is more and more widely.Pretreatment is a vital step in recognition of handwritten numerals,its strengths and weaknesses is directly related to the performance of the recognition algorithm.It describes an improved treating plan of numeral character recognition.Firstly,the method is to do the pretreatment for the input numeral character image,and then Otsu method is used to finish the global image threshold selection,through threshold processing shift grayscale image into binary image.Finally,use libsvm enhance toolbox and the method of genetic algorithm to discriminate digit.Experimental results show that the accuracy of the numeral recognition is improved.