You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I think it would be beneficial to all users if a “natural colors enhanced” RGB product could be created. After contacting EUMETSAT I provide the algorithm for that product. You can see the product hosted in EUMESAT Image Gallery (https://eumetview.eumetsat.int/mapviewer/).
The recipe for the product follows.
Kind regards
Sakis
------>RECIPE<-----
The Natural Colour Enhanced RGB (Red, Green, Blue) makes use of three solar channels: NIR1.6, VIS0.8 and VIS0.6.
It is derived from Natural Colour RGB but mitigates the effect of the unnatural cyanish colour of white snow and ice clouds by tuned gamma correction values for the principal inputs as well as by modifying the red and blue channels. In a first step a virtual cyan channel is calculated by taking the minimum of the green and blue contribution to a pixel. This channel is mapped again by a hand tuned non-liner mapping before being used for the following modification:
• in each pixel, where the red component value is less than both the green and blue value, red is increased to the minimum of the green and blue value.
• Besides, for the same pixels, the blue value is increased to the green value, if the latter is higher.
The first modification has the effect of desaturating cyan areas (i.e. making them look more greyish/whitish) while the second step reduces a greenish colour cast in the modified areas in favour of a bluish one, which appears to look more natural.
For each pixel the following is done:
• Let R, G, B be the red, green, blue colour component values of the pixel after gamma correction, i.e.:
R := gamma( albedo( NIR1.6 ) )
G := gamma( albedo( VIS0.8 ) )
B := gamma( albedo( VIS0.6 ) )
• A virtual cyan colour component is defined as:
C = min(G, B)
• Then the algorithm is implemented as follows:
if R < C then {
# desaturate cyanish pixel by raising the red component
let R := C
if B < G then {
# get rid of greenish colour cast by raising blue value to the green value
let B := G
}
}
The gamma values are manually tweaked to give a visually pleasing result, as well as an additional non-linear mapping of the virtual cyan component before using it for desaturation, as showed in the following table (Table 6), which maps values in the range from 0 to 225 (cyan) to 8-bit values (RGBA tuples). Between these stop points, linear interpolation is performed.
Dear all
I think it would be beneficial to all users if a “natural colors enhanced” RGB product could be created. After contacting EUMETSAT I provide the algorithm for that product. You can see the product hosted in EUMESAT Image Gallery (https://eumetview.eumetsat.int/mapviewer/).
The recipe for the product follows.
Kind regards
Sakis
------>RECIPE<-----
The Natural Colour Enhanced RGB (Red, Green, Blue) makes use of three solar channels: NIR1.6, VIS0.8 and VIS0.6.
It is derived from Natural Colour RGB but mitigates the effect of the unnatural cyanish colour of white snow and ice clouds by tuned gamma correction values for the principal inputs as well as by modifying the red and blue channels. In a first step a virtual cyan channel is calculated by taking the minimum of the green and blue contribution to a pixel. This channel is mapped again by a hand tuned non-liner mapping before being used for the following modification:
• in each pixel, where the red component value is less than both the green and blue value, red is increased to the minimum of the green and blue value.
• Besides, for the same pixels, the blue value is increased to the green value, if the latter is higher.
The first modification has the effect of desaturating cyan areas (i.e. making them look more greyish/whitish) while the second step reduces a greenish colour cast in the modified areas in favour of a bluish one, which appears to look more natural.
For each pixel the following is done:
• Let R, G, B be the red, green, blue colour component values of the pixel after gamma correction, i.e.:
R := gamma( albedo( NIR1.6 ) )
G := gamma( albedo( VIS0.8 ) )
B := gamma( albedo( VIS0.6 ) )
• A virtual cyan colour component is defined as:
C = min(G, B)
• Then the algorithm is implemented as follows:
if R < C then {
# desaturate cyanish pixel by raising the red component
let R := C
if B < G then {
# get rid of greenish colour cast by raising blue value to the green value
let B := G
}
}
The gamma values are manually tweaked to give a visually pleasing result, as well as an additional non-linear mapping of the virtual cyan component before using it for desaturation, as showed in the following table (Table 6), which maps values in the range from 0 to 225 (cyan) to 8-bit values (RGBA tuples). Between these stop points, linear interpolation is performed.
Input Cyan Value | Output RGBA Value
255 | 255,255,255,255
211.2 | 233,233,233,255
157.2 | 160,160,160,255
85.2 | 66,66,66,255
20 | 0,0,0,255
0 | 0,0,0,255
The text was updated successfully, but these errors were encountered: