This paper presents a simple and effective technique for converting handwritten textual character from paper document into machine readable form. The proposed method takes the scanned image of the handwritten character from paper document as input and shows the recognized character as its output. Using this method, the object in the converted binary image is segmented and is resized in a global size. After that, morphological thinning operation is applied on that resized object. The image with thinned object is then partitioned into several equal sizes of small cells. A value from each cell is estimated by calculating the proportion of the number of 1 intensity pixels and the number of 0 intensity pixels in the corresponding cell. All of these estimated values are then stored in a one dimensional array. Every element in that array is considered as a single feature value or an attribute for the corresponding image. The k-nearest neighbor classifier is used to classify the handwritten character into the recognized classes of characters. Feature values are estimated from training example images and the classifier is trained using the attributes. After training attribute values for sample image are extracted and passed as inputs in the k-nearest neighbor classifier and the sample image object is grouped using the training dataset into the desired character classes. The proposed technique takes less time to compute, has less complexity and shows desired performance in matching the handwritten characters with the machine readable form and in recognizing them.

Character Recognition Morphological Thinning Operation Cell Feature Value K-nearest neighbor classifier