A Machine Editable Format for Kannada Handwritten Character Recognition Using A Dissect and Connect Technique
H. N. Srikanta Prakash · 2014
HCR for Indian Languages is a challenging task where there is relatively little work has been done. The typical nature of the kagunita , the compound characters in which one or more consonants combine with vowel, increases the complexity of the character for recognition. The segmentation is the major task of any OCR. The aim is to provide segmentation techniques that have been developed, the classical approach consists of methods that partition the input image into sub-images, which are then classified. The operation of attempting to decompose the image into classifiable units is called dissection. We try to reduce the character complexity by segmenting each character into four sub images namely main unit, right adjunct, bot- tom adjunct, top adjunct so that we can decrease the number of symbols for the classification. In this paper, we experiment with the use of moments features on Kannada kagunita. Moments and statistical features are extracted from the cut images. These subspace features are used for recognition on Neural Network Classifier. The result is obtained in the editable format which can be edited for the future purpose.