Handwritten Character Analysis

challenging and subject of much attention in the field of recognition. This is due to the fact that styles of people vary to a great extent and it becomes difficult for the computer to recognize the handwritten characters. Various techniques are proposed in literature including restrictions like specific writing styles-uppercase or lowercase characters. A more difficult problem is the recognition of characters when the writing style is not known a priori. This is a very relevant problem because most of the practical applications do not give any hint about the writing style and hence the recognition system itself has to find out and manage different writing styles. The paper discusses some of the soft computing techniques for the Handwritten

As shown in figure 1, first input samples are passed through a scanner to the system, where preprocessing converts the image into a form suitable for subsequent processing and feature extraction. Next stage is segmentation, where the input image is segmented into individual glyphs. This step separates out sentences from text and subsequently words and letters from sentences. Feature extraction black, extracts important features that forms a vital part of the recognition process. During training such data is stored in the database where, during classification, a character is placed in the appropriate class to which it belongs. Handwritten Character Recognition has been a challenging research domain due to its diverse applicable environment [3]. Handwriting, as has always been, is assumed to be continued as preferred means of communication. Effective HCR systems need to be designed to convert these handwritten documents into an editable format. HCR systems aim at higher accuracy, with considerably reduced computational and storage space requirements. In [4], Rajbala et. al have discussed various types of classification of feature extraction methods like statistical feature based methods, structural feature based methods etc. Tirthraj Dash et al have discussed HCR using associative memory net (AMN) [5]. In Zoning-based classification [6] a membership function defines the way a feature

HOG being a gradient –based descriptor is stable on illumination variation and hence promises to be very efficient feature descriptor for handwritten digits. The paper [9], heuristic character recognition is explored using a randomized algorithm. Heuristic is used to generate the FCC correctly to represent the characters. SVM classifier is used to recognize the characters.

Our current work focuses on comparison of various approaches for offline handwritten English alphabets and numerals and rest of the paper is organized as follows. In section 2, Bi-directional Associative Memory is discussed. Section 3 Neural Network model is presented, Section 4 shows simulation results and finally, Section 5 concludes the review. bi-directional Associative memory