The frequency of some image signal was traditionally given by a number of cycles-per-millimeter (or per other spatial unit). This is a historical unit of measurement that is related to the invention of photography and comparing lenses.
A cycle is also called a pair of lines. It is created by the alternance of a dark-then-white pattern. If the black and white lines have the same width, then you have a pure frequency. If the lines get thinner and thinner (as in most resolution test charts), then you have some chirp.
A constant region, whatever the color, is some continuous signal. High frequencies correspond to spatially fast changes in intensity: a pattern of dots or of thin zebras is high frequencies.
Since an image is a 2D signal, frequencies are defined with respect to their horizontal and vertical components (obviously, some diagonal pattern can be created by having non-continuous horizontal and vertical components).
You can find an example of resolution chart here.
If you want to experiment by yourself, you can try to generate various frequencies by creating fake 2D Fourier transforms with only 1 non-zero frequency and invert them.