Research And Hardware Implementation Of Key Algorithms In Fpga-based Image Processing System

Image preprocessing system is developing towards high-speed, high-resolution, high-integration and high-reliability. To be concerned with software processing systems'inefficiency of speed and real-time performance, using hardware to realize image processing algorithms tend to be a new research focus. FPGA(Field Programmable Gate Array) is one of the most widely used programmable logic devices. FPGA can be used to realize hardware parallel and pipeline processing technology, and has the advantage of fast processing speed and nice transplanting property, thus FPGA is the best choice for image preprocessing which features simple algorithms and enormous data rate.Domestic researches on FPGA based image processing systems have been widely developing home and abroad. But compared with foreign research, there is a big gap in real-time processing of color image, hardware implementation of advanced image processing algorithms and optimization of processing algorithms etc. Under this background, analysis and research have been made about real-time preprocessing algorithms of color image, and some improvements have been proposed in designing image processing algorithms which are suitable to be realized on hardware. Then by using system interface control and hardware resource distribution technology, the proposed algorithms have been realized on the DE2 hardware developing board which takes Cyclone II FPGA as core processor. Finally after analyzing signal flow and interface control, the proposed image preprocessing system is transplanted into a real-time image acquisition system. And the whole system achieves the ability of real-time acquisition, storing, preprocessing and display for video images.In chapter 1, after studying plenty of relevant research documents, some image preprocessing algorithms are chosen, due to their nice suitability of hardware implementation, to realize an image processing system on FPGA-based hardware platform.In chapter2, two CAD tools, MATLAB 7.0 and QUARTUSII 9.0, are introduced. Steps are presented to show how to design image processing algorithms and make function simulation in MATLAB 7.0 software; Developing flow and design methods are detailed to show how to implement an image processing system in FPGA; VERILOG hardware programming language is also talked about. In chapter 3 and 4, the most commonly used two image preprocessing algorithms are researched with emphasis: denoising and sharpening. Detailed analysis and research are made on space domain image processing algorithms and their implementation scheme. To be concerned with median filtering, innovative improvements have been proposed that using transversely extended window as filtering panel, which can efficiently raise the operating speed of median filtering. Finally median filter and laplacian filter are implemented on FPGA hardware, using image parallel transformation and pipeline processing technology, and these filters are designed as VERILOG HDL modules.In chapter 5, interface control has been studied about image acquisition system and image processing module, according to which image processing module is transplanted into real-time image acquisition system. The image acquisition and preprocessing system has been tested in real-time sunning state, and its real-time performance and power consumption are also analyzed. Finally the sixth chapter summarizes this thesis research and forecasts the prospect of FPGA-based image processing system development.