Finding bilateral symmetry axes in almost-symmetric images

Although the human visual system is adept at finding axes of bilateral symmetry in image data, the prevailing computational method for finding symmetry axes, proposed by Marola (1989), was found to be unreliable for large immunohistological images. A modification of the Marola method which exploits the current understanding of how the human visual system processes symmetry is proposed and tested. This modification involves using only the portion of the data within 1 degree of the visual angle of the symmetry axis. Symmetry axes found using the new method were judged by inspection to be better than those found using Marola's method.