Preprocessing phase for handwritten devanagiri word recognition

In this paper we focused the importance of the pattern classification and its application. We list the characteristics of Hindi language writing style, furthermore focused onthe preprocessing step of the recognition system. We describe a complete system for the recognition of isolated handwritten Devanagari word using Fourier Descriptor and Hidden-Markov Model (HMM). The HMM has the property that its states are not defined as a priory information, but are determined automatically based on a database of handwritten word images. In this work the image database consist of 500 images of handwritten Devanagari words from 50 different writers. Before extracting the features, the images are normalized using image isometrics such as translation, rotation and scaling. After normalization the Fourier features are extracted using Fourier Descriptor. An automatic system trained 500 images of image database and word model form with multivariate Gaussian state conditional distribution. A separate set of 100 words was used to test the system. The recognition accuracy for individual words varies from 90% to 98% for number of states per model N=3.