The FPGA Realization Of SURF Image Mosaic Algorithm

Image mosaic technology is a hot research topic in digital image processing. In recent years, image mosaic technology has been well applied to many fields such as machine vision, unmanned aerial vehicle and vehicle monitoring with the maturity of the technology. Image mosaic technology based on feature, the panorama is perfect and not affected by light, rotation, is the current research focus of image mosaic.In this article, we achieve the whole image mosaic algorithm, based on study and learning Speeded up Robust Features (SURF) algorithm, Random Sample and Con-sensus (RANSAC) algorithm and the Bilinear Interpolation algorithm. The realization of the whole image mosaic algorithm includes SURF algorithm, interest point match-ing module, transformation matrix module and image panorama module.Firstly, we developthe SURF algorithm using the MATLAB software until the two images mo-saicking one image that consisting to our expectation. Secondly we develop the algo-rithm using the FPGA Advantage. In the process of development, we write the MATLAB program according to the FPGA frame. Finaly, we perform many simulation to detect the image mosaicking. The results are consistent to the MATLAB program.The realization of the SURF algorithm includes the interest point solving, the main direction of interest point solving, the interest point descriptor solving. Image mosaic technology is more complex, in order to realize this mosaic algorithm, the implementation platform being DSP, ARM and FPGA. However, the FPGA has the processing ability of the DSP, and embedded within two ARMs, as long as realization design, we can achieve the mosaic algorithm using FPGA. Firstly, we use MATLAB to develop the SURF algorithm. Secondly, we use FPGA Advantage to design the algo-rithm, using Quartus ? to synthesize, using Modelsim simulation. Finally, we get the Panorama.