Rapid FPGA prototyping of Gabor-wavelet transform for applications in motion detection

摘要

Many motion detection algorithms have been developed for various applications such as machine vision, robotics and content based coding techniques. In this paper, a new method for motion estimation is evaluated with the goal of tracking human face in video stream. Gabor wavelet transform is used to overcome the aperture problem in image flow estimation. A histogram over this image flow provides a motion hypothesis, which can be used to detect the motion information in image sequences accurately. The algorithm is implemented in FPGA in order to speed up the computation of Gabor wavelet transform. System level approach to the design of FPGA (Field Programmable Gate Array) for prototyping is described. Although FPGA does not offer an optimized hardware implementation when compared to ASIC (Application Specific Integrated circuit), it allows short development time and enables verification of algorithms in hardware at low cost.