Mach bands

In a series of areas of different gray coloration that have no color grading in itself, we observe along the borders Mach strip ( after the physicist Ernst Mach, 1865). It is light and dark stripes that enhance the contrast between the areas.

A basal mechanism for the processing of visual impressions in humans is the identification of lines and edges. We see often faster than photos that contain only shades of color line drawings. The canals of Mars are an example of a recognition of linear structures, where not really exist, but at best short line segments or shadow effects.

Already processing in mind emphasizes contrasts and lines, because the retina correlates the irritation of adjacent photoreceptor cells into the brain even before forwarding. Although in the right figure, the upper surfaces possess only discrete shades of gray, we look at the transitions of a brightness curve, the excessive contrast: The dark edge appears darker, the bright area brighter. The gray scale in the image below appears inhomogeneous, although the gray value changes linearly from left to right.

The cause of this phenomenon lies in the performance of connection of the receptors in the retina. Approx. 100 million receptors control around 1 million receptive fields that provide the signals for processing. By wiring the receptors not only amplify signals, but they can also slow down depending on the origin (lateral inhibition).

The principle can be best illustrated by considering four receptors. On the receptors A and B falls higher intensity light, in the example image with the intensity of 30 on the receptors C and D low intensity light 10 The excited receptors inhibit the two respective neighboring receptors with a fixed proportion of the amount that they get themselves in excitement (in the example with one-tenth ). Thus, for the nachgeschaltenen neurons following excitation distribution:

  • Receptor A: 30 -30 -30 * 0.1 * 0.1 = 30 -3 -3 = 24
  • Receptor B: 30 -30 -10 * 0.1 * 0.1 = 30 -3 -1 = 26
  • Receptor C: 10 -30 -10 * 0.1 * 0.1 = 10 -3 -1 = 6
  • Receptor D: 10 -10 -10 * 0.1 * 0.1 = 10 -1 -1 = 8

Thus the signal is inhibited more strongly or less strongly to the boundaries of two different bright areas and there are the do 's fringes.

The effect occurs, inter alia, in computer graphics for calculated, illuminated surfaces (eg when Gouraud shading ).

The image processing in the brain also supports the edges perception, see for example the Kanizsa Triangle.