Telugu Word Image Recognition and Retrieval using Non-Sub Sampled Contourlet Transform
Keywords:
Image database, non-sub sampled contourlet transform, non-sub sampled pyramit filter bank, non-sub sampled directional filter bank, texture classificationAbstract
There has been increasing focus and research on Telugu word image recognition and word retrieval in recent years. This has direct applications in advanced technologies like Artificial Intelligence, Public addressing systems, and some commercial personalized systems. There are limitations and complexities in recognizing a Telugu word image due to lack of abundant technical literature and sophisticated optimal methodologies. A general method of identifying a word image is by tracing out its relevant word in the Telugu Word Images Database based on extracted image features and similarity measurements. This paper explains the procedure adopted in Non-sub Sampled Contourlet Transform for extracting features of a Telugu word image,measuring its similarity with an image stored in database and retrieving a relevant word from the database. The NSCT is efficient in relevant word retrieval in comparison with Non-NSCT methods.
