Design and Implementation of Automated Image Handwriting Sentences Recognition using Hybrid Techniques on FPGA

Abstract

The validation of documents such as recognition of optical character, the sign which is written by hand are the main drawbacks involved in the identification of human and their addresses, codes of the post written on the envelops, manuscript evaluation, understanding the transactions of money and documents of the bank that are written in the English language. The conceptual model was written by hand for the real-time application that deals with the handwritten identification enables a comprehensive computerized system to identify the data written by hand which is more efficient and is free from noise. The proposed framework consists of filters based on Probabilistic Patch (PPB), identification, and analysis of the Canny edge. With the application of a Probabilistic Patch-based filter, the recursive speckle noise and additive Gaussian noise are processed. The words in the document are obtained by using the structure of Lifting transformation, the edges of the word are identified with help of Canny edge recognition. At last, the database validates the text as correct or incorrect. With the application of the Embedded Development Kit (EDK) and Software Development Kit (SDK), the entire framework is developed. The hardware used is in this work is Virtex-5 FPGA board which is the integration of SDK and EDK with XC5VLX50T as the part name.