| With the development of LSIC and microelectronics technologies, digital imageprocessing techniques are applied to social life and the military field more and morewidely, and numerous domestic and overseas IC corporations have produced somespecial-purpose image processing chips. However, in other special application fields,chips with dedicated functions have not been designed yet, so the desired functions canbe achieved only through other methods. In this dissertation, a real-time imageprocessing architecture with FPGA as the crucial processing chip is proposed. It is usedto achieve real-time fusion of two channels of high definition DVI video signals. Thetwo channels of video signals are processed by the DVI decoding chip which isproduced by TI Corporation, which makes the TMDS signal converted into the RGBsignal. And then the EP3C15F484C6chip designed by ALTERA Corporation is utilizedto accomplish real-time preprocessing and fusion of images. Next, the signal is output tothe DVI decoding chip or the D/A conversion chip so as to get the digital DVI signal orthe analog VGA signal. This design focuses on the following:(1) The hardware circuit of DVI video signal fusion system is designed. Thecircuit includes DVI decoding circuit, DVI coding circuit, D/A conversion circuit,UART communication circuit, FPGA, memory circuit and so on. In the process ofcircuit design, according to the direct disturbance caused by the high-frequency signal,dissertation will focus on circuit layout and treatment of the power supply system.(2) Verilog program corresponding to the peripheral circuit of the system isdesigned, and program debugging on the host computer finished. These designs includethe design of VGA controller module, designs of the resolution identifying module, thevideo polarity discriminating module, the serial port communication module and theSDRAM controller module etc.(3) The real-time image processing algorithm applied to the video fusion system isintensively analysed. Content of the analysis includes the relative planar motionalgorithm of an image, the image interpolation and magnification algorithm, and the real-time fusion algorithm of two-channel video signals and so on. In addition, theMatlab simulation graph and the Modelsim functional simulation graph relevant to thesealgorithms are presented as well as the final hardware testing chart.The system has the maximum resolution of1920×1080and the maximumoperating frequency of162MHz. It can support input with different resolutions, alsosupport the top graph and the underlying graph of any video image. The system is notonly stable, the image quality is also good. With high-quality images, this system worksstably and has been utilized by a research institute in Chengdu. |
