Click on the image for a medium-size version (640 x 480, ~170 kbytes)
Contrast sensitivity is a measure of the limit of visibility for low contrast patterns -- how faded or washed out can images be before they become indistinguishable from a uniform field? (Think of driving in a fog). It is a function of the size (coarse/fineness) of image features, or the spatial frequency. The test image shown above was first produced by Campbell and Robson (1968; ref. 1) to illustrate the form of the function in a very intuitive manner -- using everyone's own visual system and without time-consuming measurements.
In this image, the luminance of pixels is modulated sinusoidally along the horizontal dimension. The frequency of modulation (spatial frequency) increases logarithmically, i.e., with exponential increase in frequency from left to right. The contrast also varies logarithmically from 100% to about 0.5% [or whatever your 8-bit gray scale display gives you] (bottom to top).
The luminance of peaks and troughs remains constant along a given horizontal path through the image. Therefore, if the detection of contrast is dictated solely by image contrast, the alternating bright and dark bars should appear to have equal height everywhere in the image. However, the bars appear taller in the middle of the image than at the sides. This inverted-U shaped envelope of visibility is your contrast sensitivity function. The exact location of the peak depends on the viewing distance. Try moving farther away from the display, and back closer. Note that the apparent position of the peak shifts as you do this. Therefore, the inverted-U shaped envelope is not in the image, but reflects the property of your visual system.