Symmetry detection is important in the area of computer vision. A simp
le and fast algorithm for the recovery of rotational symmetry paramete
rs is developed. The fold number of a rotationally symmetrical shape i
s obtained from the gradient orientation histogram of the input gray-l
evel image using the Fourier method, based on the relationship between
the periodicity of the orientation histogram and the fold number of t
he symmetrical shape. It takes only 0.06 a CPU time for a 256 x 256 gr
ay image on a Sun SPARCIO. Both simulated and real images are tested a
nd the results are very convincing. (C) 1997 Society of Photo-Optical
Instrumentation Engineers.