An entire optimised SIFT algorithm has been proposed for FPGA based real-time image matching applications in this paper. Not only has the corresponding software model been built up, but the key part of the optimized SIFT algorithm,which is the SIFT feature detection module, has also been implemented on a Xilinx Virtex-5 FPGA. It has been confirmed from the standard test that our SIFT feature detection module is able to extract features from a typical image of 640x480 pixels with 31 ms, which is faster than any existing systems. It can be seen from the matching performance comparison that our optimized SIFT algorithm has achieved a similar level to SURF, which is one of the well recognized image matching algorithms.
1