Handwritten Gurmukhi Numeral Recognition using Zone-based Hybrid Feature Extraction Techniques

This paper presents an overview of Feature Extraction techniques for off-line recognition of isolated Gurumukhi numerals/characters. Selection of Feature Extraction method is probably the single most important factor in achieving high performance in pattern recognition. Our paper presents Zone based hybrid approach which is the combination of image centroid zone and zone centroid zone of numeral/character image. In image centroid zone character is divided into n equal zone and then image centroid and the average distance from character centroid to each zones/grid/boxes present in image is calculated. Similarly, in zone centroid zone character image is divided into n equal zones and centroid of each zones/boxes/grid and average distance from zone centroid to each pixel present in block/zone/grid is calculated. SVM for subsequent classifier and recognition purpose. Obtaining 99. 73% recognition accuracy.