Script Identification from Handwritten document Images Using LBP Technique at Block level

摘要

The documents are in multilingual form in India, it is required to automatically identify type of the script and feed script document to the appropriate Optical Character Recognition system for information retrieval. This paper presents an efficient handwritten script recognition method using Local Binary Pattern operator. The features are extracted from a block of handwritten document image. Recognition of the script type is done using Nearest Neighbor and Support Vector Machine classifiers. Experiments are performed on images of handwritten documents written in English, Hindi, Kannada, Malayalam, Telugu, and Urdu scripts. KNN and SVM classifiers yielded recognition accuracy of 98.46% and 99.5%, respectively.