In a picture file, colours are coded as the intensities of the three colour components red, green and blue. These codes are on a scale from 0 to 255 representing the brightnesses. Hemisfer can display a graph (so called histogram) showing which proportion of the pixels in the picture have any brightness. The brightnesses are thus on the X axis, either as coded brightnesses or as light intensities according to the γ (gamma) value of the picture. Moving the mouse over the graph let the brightness being displayed beside the mouse cursor (tool tip). The frequencies on the Y axis are represented either on a linear or on a logarithmic scale. There are five curves on this graph: three for the single colour components (red, green and blue), a white curve for their sum, and a yellow curve for the current weighting of colours. These frequencies are calculated within the analysis rings. Note that pictures in 256 grey tones show only the white curve, while pictures in 256 colours will show quite erratic frequency curves.
When pictures are taken under an overcast sky, the colour channels have usually similar curves. When the sky is blue, then the second blue peak comes at higher brightness values than other colours. Displaying the colour frequencies may help to decide which colour channels to consider in the analysis of the picture.
The colour frequencies can mainly help to understand problems arising in setting the threshold. If set, the threshold is displayed as a vertical line on the graph. If the calculation was done according to Nobis & Hunziker (2005), then the curve of average gradients is also displayed. In an ideal case, this curve shows a clear maximum corresponding to the threshold, which divides the pixels in two distinct peaks: the dark ones (canopy) on the left and the light ones (sky) on the right. If the calculated threshold is not satisfactory, check first if a secondary peak of the gradients curve may be better. Clicking on the graph is an easy way to set a manual threshold. If the picture in the main window is in black and white, then the effect of a new threshold will be displayed automatically.