The term image is used extensively in visual art and visual science, but its application has often been so catholic as to be counterproductive. Images proliferate in our environment and in our language. One of the attractions of the term is its appeal to the spatial dimension both in pictures and in their internal representations, as well as in describing them verbally. However, there is no consensus concerning the definition of an image. In fact, its meaning is typically treated as self-evident. Scientists use it in quite different ways to artists. Vision in humans and image processing in machines is considered to proceed from the analysis of two-dimensional images formed either on the retina or in a camera. It is unlikely that human vision (or that in other species) starts from a series of single time-frozen projections of the world. This preoccupation with the retinal image reflects a confusion between the optical projection onto a surface and the biological processes that are initiated by light. This, in turn, relates to the historical contrast between optics and observation, between light and sight. These distinctions can be made more explicit with a model of perception that involves a sequence of imaging stages. Pictorial images provide allusions to spaces they do not occupy; they refer indirectly to the objects they represent. The term image is used repeatedly in the model, but always with some qualification, as in retinal image, visual image and mental image. Images in art refer, in the first instance, to the marks made on a surface. The term image is also used to convey the impact that the pictorial image has on the viewer: this impact can refer to its allusory three-dimensionality (which I refer to as the graphical image). Many graphical examples of visual allusions will be presented.