Finding the closest color
Author
Ok, so I''ve got this array of 256 colors (type RGBQUAD).
What algorithm should I use to, in this array, find the absolutely closest color to ANY color?
Thanks for your time!
/mr BiCEPS
well, if you know the number of the color you''re looking for (0-256) or whatever, you can just look for a matching number. you could do this by creating a look-up table or something.
JoeMont001@aol.com
www.polarisoft.n3.net
JoeMont001@aol.com
www.polarisoft.n3.net
RGBQUAD Colors[256];int FindClosest(RGBQUAD rgb){ int iIndexFound = -1; int iDifferenceFound = 100000; for(int i=0; i<256; i++) { int iDifference = abs(Colors<i>.rgbRed - rgb.rgbRed) + abs(Colors[i].rgbGreen - rgb.rgbGreen) + abs(Colors[i].rgbBlue - rgb.rgbBlue); if (iDifference < iDifferenceFound) { iIndexFound = i; iDifferenceFound = iDifference; } } return iIndexFound;} 
Hmm, why does this algorithm looks familiar?
Sludge Software
www.sludgesoft.com
Developing a secret of mana style role-playing-game
Sludge Software
www.sludgesoft.com
Developing a secret of mana style role-playing-game
Ah, yes, Baskuenen''s code finds the index of the color with the same *brightness* ,not necessarily the same colour.
(I KNOW, I had this problem too)
FReY
(I KNOW, I had this problem too)
FReY
Baskuenen''s code is calculating the Manhattan distance (it''s really called like that) of two colors. Another way would be to use Euclidian distance ... sqrt((r1-r2)^2 + (g1-g2)^2 +(b1-b2)^2)
The problem is ... there is no ''correct'' way. Both ways of calculating the distance are valid. You might even go further and really use the brightness weighting of the RGB components to get better results:
distance = 0.3 * abs(r1-r2) + 0.6 * abs(g1-g2) + 0.1 * abs(b1-b2)
In the end, whatever works is good ... not a good answer, but so very close to the truth.
MK42
The problem is ... there is no ''correct'' way. Both ways of calculating the distance are valid. You might even go further and really use the brightness weighting of the RGB components to get better results:
distance = 0.3 * abs(r1-r2) + 0.6 * abs(g1-g2) + 0.1 * abs(b1-b2)
In the end, whatever works is good ... not a good answer, but so very close to the truth.
MK42
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement