The Implementation Of Image Matching System Based On Embedded Processor

Along with the rapid development of science, the image matching has become a very important technique in image information processing realm. It not only is the key technique in image measuring system, but also is the most basic method in target fixed position. Although great development has been made in image matching arithmetic, many only stay on the paper limited by hardware condition. How to optimize arithmetic to make it has the application value is a very meaningful research work.The thesis developed an image matching device based on "DSP+FPGA" hardware structure on which realized Phase Only Correlation (POC) arithmetic and Rotate Invariability Phase Only Correlation (RIPOC) arithmetic. The device not only can detect the shifted image, also can detect the rotated image. Using the the log-polar transformation, we can converse the revolved angle problem to be an easy shifted postion question.The most decisive works lie in the control of the image-capturing timing and the optimization of the image-matching arithmetic. In our system, the camera works under the random trig mode, we must extract the related synchronization signal from the captured video signal and conrol the trig and synchronization of camera, all this work can be completed by FPGA. The complicated FFT/IFFT operation is also done by FPGA because of its parallel and pipeline character.This kind of hardware structure not only can realize complex arithmetic also can get the maximal processing speed. The last examination result proved that this project has stronger flexible, general and real time. It can detect the shifted and rotated image quickly and accuracy above 99%.