When a moving object is recorded in a sequence of digital image frames, the Fourier transform may be used to detect its motion and speed. This technique is especially suitable for detecting a slow movement in a noisy environment. When detecting a slow movement, a large number of frames may be needed, which involves a large amount of samples. If frames are skipped uniformly to reduce the number of samples, aliases from higher frequencies may occur. Random sampling at a sub-Nyquist rate can overcome aliasing, which fits this application perfectly if the frames are taken irregularly in time. Since frames are usually taken regularly, a pseudo-random method may be adopted.
展开▼