An Approach to Recognise Odia Characters by Extracting Features from Fragmented Quadrants

The system introduced in this study concentrates on the recognition of Odia characters. The input to the system is an image in .jpg format which has been created by scanning a document or converting a document to an image by software. The document consists of all Odia alphabets written using the font family 'AkrutiOriAshok-99' in a particular font size in bold style. The font sizes that have been considered are 18, 20, 22, 24, 26, 28, 36, 48, and 72. The noise is removed and line texts are segmented from the input image. Then individual alphabets are segmented from a line text using the Character_Segment part of the system proposed in this study. Then the features are extracted from the segmented individual alphabet and the features are searched in the dictionary of features stored in an excel file for a correct match one by one. The proposed system is also responsible for creating the dictionary of features of all Odia alphabets. Hence, the system in this study has four parts, Segment_Line, Character_Segment, Build_Dictionary, and Recognise_Character. According to the testing results, the system proposed in this study has achieved a correctness accuracy of recognition of Odia alphabets of 99.8%.