Unconstrained Handwritten Kannada Word Recognition Using ESDCW Method

Segmentation is crucial in the Human Character Recognition System for extracting text lines, words, and characters from handwritten Kannada documents. The proposed system uses the Enhanced Skew Detection and Correction for Words (ESDCW) algorithm to extract text lines, words, and characters, and corrects skew lines using skew angle repetition. Preprocessing methods include filtering, gray scale conversion, and binarization. The system recommends preprocessing, dilation, labeling, deskewing, and inserting words into the new image. Unwanted information is removed using the bounding box technique, avoiding overlapping words. The system achieved an average segmentation rate of 96.38% on fully unconstrained handwritten Kannada documents.