Radial symmetry is an important perceptual cue for the feature-based representation, fixation, and description of large-scale data sets. A new approach based on iterative voting along the gradient direction is introduced for inferring the center of mass for objects demonstrating radial symmetries that are not limited to convex geometries. The kernel topography is unique in that it votes for the most likely set of grid points where the center of mass may be located. Initially, it is applied in the direction of the gradient and then reoriented iteratively in the most probable direction. This technique can detect perceptual symmetries, has an excellent noise immunity, and is shown to be tolerant to moderate perturbation in scale. Applications of this approach to blobs with incomplete and noisy boundaries, multimedia scenes, and scientific images are demonstrated.