Dissertation
Dissertation > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device

Based on FPGA and DSP Text Image Acquisition and Locate Studies

Author ChenYan
Tutor ZhaoZuoZuo
School Liaoning University of Science and Technology
Course Control Theory and Control Engineering
Keywords DSP FPGA Text Location Image Processing Mathematic Morphology
CLC TP391.41
Type Master's thesis
Year 2012
Downloads 80
Quotes 0
Download Dissertation

In this paper, the text image acquisition and location was studied, including acquisition and location of two parts. The first part is based DSP and FPGA is researched. The special structure and optimal property of DSP add the high speed performance and flexibility of FPGA, which can easily solve the contradiction between mass computation quantity and real time procession. It also simplifies peripheral circuit and lowers the degree of design difficulty. In the system, as a core module of image acquisition, FPGA is responsible for collecting signal timing control, the input digital signal conversion framing format, and set up dual-port RAM and DSP real-time video data interchange; Application of DSP peripheral McBSP(multichannel buffered serial port to receive data ), DMA to transmit data, real-time collection requirements.Also designed the video image text location algorithm, the first based on Sobel operator into the gradient algorithm for image edge detection, which quickly and accurately find the image edge. After processing the text using mathematical morphology edge image. Then apply the region growing algorithm to locate the target. Finally, according to the text of the characteristics of the text area to discriminate, to determine the text area. And simulation with MATLAB, and gives the final result. Experimental results show that the proposed method is not only simple, and can more accurately locate the video text, text area can effectively remove from the background image.

Related Dissertations
More Dissertations