next up previous
Next: Background Summary Up: Background Previous: Interactive Techniques

Perception Studies

 In computer graphics there has been very little work on quantifying claims like ``Image 1 conveys more shape information than Image 2.'' However, perceptual psychologists have performed numerous studies and experiments, trying to learn about human recognition and the way we visually process information. Perception studies can help to provide a quantitative analysis instead of merely giving an educated hypothesis on the effectiveness of an image to convey shape information. Studies by Christou et al. [7], Braje et al. [4], and Biederman et al. [2] support the use of line drawings as an effective method for providing substantial shape information.

A study by Christou et al. [7] showed four scenes to subjects. Each scene was composed of a number of planar, cylindrical, and ellipsoidal surfaces. One scene contained shaded surfaces (similar to Phong shading); another scene with textured objects; a scene which only included contours, the line-drawn edges of the objects; and a scene with contours and textured objects. The subjects were asked to specify the surface attitude, the orientation of the local tangent plane at a point on a surface with respect to the viewing direction, at random points on in the scene. These tests showed that the subjects had improved performance in the scenes containing contours. They concluded, ``a few simple lines defining the outline of an object suffice to determine its 3-D structure. The ecological significance of contours is clear. They delineate the different components of complex objects and the different parts of a scene''(p. 712).

Another recognition study by Braje et al. [4] found that humans fail to use much of the information available to an ideal observer. Their conclusion was that human vision is designed to extract image features, such as contours, that enhance recognition, disregarding most of the other information available.

Biederman et al. [2] concluded that simple line drawings can be identified about as quickly and as accurately as fully detailed, textured, colored photos of the same object with the same viewpoint. The question they tried to answer was whether the presence of gradients made it easier to determine an object's 3D structure over that which can be derived solely by the depiction of an object's edges. They concluded, for example, that one could determine the curvature of a cylinder, planarity of a square, or volumetric characteristics of a nonsense object from a line drawing, without the presence of surface gradients. They noted that instruction materials for assembling equipment are more easily followed when the parts are drawn instead of photographed. Their opinion is that reproduced photographic images typically have insufficient contrast for determining the contours of components. Although it seems that one could modify a photograph to get the necessary contrast, there are other techniques, like cut-a-ways, that cannot be easily accomplished, if at all, with photography.


next up previous
Next: Background Summary Up: Background Previous: Interactive Techniques