Original Image. One of the four random images from the experiment's training session, with various levels of difficulty.
Color Lens. The enhanced image after the feature has been found with the Color Lens. There are clear artefacts, but these are easily interpreted as overexposed and underexposed regions.
Histogram Equalization. Uses KamLex's PhotoShop plugin for image equalization. The original brightness distribution is gaussian, histogram equalization cannot do better than flattening it a little bit. This increases the dynamic range of gray regions but decreases the dynamic range of dark and bright areas. The feature on the 1st example is more visible but the one on the 3rd example (dark background) becomes invisible.
Tone Mapping. Uses HDRSoft's PhotoShop plugin for tone mapping with default settings. Tone mapping increases contrast locally. Its effects are soft (they are supposed to look natural on photographs) but it makes the feature sligthly more visible on some low-density noise images.
Local Hist. Equal. (256 pixels). Uses Alexander Belousov's PhotoShop plugin for multiple histogram equalization with a tile size of 256 pixels. Multiple histogram equalization also increases contrast locally, but compared to tone mapping this thing can be much stronger. It reveals features (especially on low-density noise) but all large-scale information is gone.
Low Pass. Uses Photoshop's high-pass filter with a radius of 4 pixels + automatic levels. This is local contrast enhancement pushed to the extreme. Obviously the feature is clearly visible, but there is little left from the original image.
