Implementation of Real Time Retrieval of Text from Camera/Scanned Document Image using FPGA

Abstract

For the few decades, the work on computer captured/scanned-skewed document image is a major challenge for the researchers in the field of OCR. Imaging the old historic or bulky book causes skewness and distortion on the image. The proposed metric retrieval method helps in preprocessing, skeletonization, skew detection, skew correction and recognition using horizontal and vertical text line extraction of the image. Text line detection depends on the goal of digitization. Digitization is helpful in museum and library where old historic and damaged manuscripts are imaged. In the real time the image is captured using the camera fitted in FPGA kit (cyclone II). The captured image is preprocessed with the verilog code which is converted from MATLAB code using simulink HDL coder.