Design and FPGA Implementation of DWT, Image Text Extraction Technique
Adesh Kumar, Prakshi Rastogi, Pragyan Srivastava · Procedia Computer Science · 2015
In digital image processing, detection and extraction of text from a documentary image is found a challenging task, especially for inclined, vertical and circular text. In the extraction method, the image is segmental in text regions from a compound image. There are many techniques to segment the image but HAAR DWT is one of the popular and fastest methods, because its coefficients are either 1 or -1. HAAR DWT detects 3 kinds of edges for colour images and removes non texted regions. The paper focuses on the design, modeling and simulation of the proposed method with the help of VHDL programming language. The design will be compared with the existing methods, and the speed and existing hardware is optimized in terms of memory and time parameters. The design is developed in Xilinx 14.2 ISE software and is synthesized on Virtex -5 FPGA.