Abstract
Purpose: :
For patients with advanced retinitis pigmentosa, glaucoma or a retinal prosthesis, only a small piece of the environment can be visually perceived at any time. An understanding of how pieces of a visual scene are put together to form a global percept is important for low vision evaluation and rehabilitation. A polygon of random vertices is a simple abstraction of an unfamiliar visual object. It can be displayed by vertices, sides or pieces of different sizes. Normal human subjects’ ability to use pieces of a polygon to perform symmetry judgment was evaluated.
Methods: :
Random polygon outlines were generated. Each polygon had 10 vertices, was 300x300 pixels in size (8.5 deg visual angle at 60 cm), and had a line width of 3 pixels. The polygons had different degrees of axial symmetry and random symmetry axis. A polygon was displayed in several ways, vertices one-by-one, sides one-by-one, and 4, 9, 16, 25 or 36 pieces one-by-one. Each part was presented 100 msec. Parts could be displayed in sequential or random order. Symmetry axis could be hidden or shown to the subject before each trial. In an experimental session, polygons of good symmetry were mixed with asymmetric polygons. The subject’s task was to determine if a polygon was symmetric or not. Eye movements were not restricted.
Results: :
Normal subjects could judgment symmetry of a polygon from its pieces, even when the polygon was cut into 25 or 36 pieces. A 50 to100-trial practice might be needed to reach an 80-90% correct rate. Presenting the pieces in random order and hiding symmetry axis had little impact on performance. Random jittering of piece positions up to 16 pixels had only moderate effect on symmetry judgment. Symmetry judgment was accurate if sides of a polygon were presented in sequential order, even without knowledge of symmetry axis, but was impossible if sides were presented in random order, with or without symmetry axis. A 4-pixel random jittering of side positions had little effect, but a larger jittering impaired performance. Symmetry judgment was accurate if isolated vertices of a polygon were presented in sequential order and with knowledge of the symmetry axis, but was not possible without knowledge of the axis or when vertices were presented in random order.
Conclusions: :
Perceiving global characteristics of a polygon from pieces cannot be explained simply by retinotopic or spatiotopic painting of the retina or visual memory. At least for the task of symmetry judgment, human subjects use features that are more complex than vertices or lines, typically corners made of two or more lines or three or more vertices, either presented simultaneously or sequentially. Visual field sizes large enough to contain such features are crucial for low vision patients to perceive large objects.
Keywords: pattern vision • shape, form, contour, object perception