| Author |
 Topic  |
|
|
Dudskoni
4 Posts |
 Posted - Oct 28 2012 : 15:21:35
|
I am using the following function to find the difference between two colors defined in Lab. The problem is that I am getting intermittently the error “Invalid floating point operation” error when using this function. I am sure there is no sqrt of negative number or division by zero. What is going on? Is there a better way of doing this using Imageen?
thanks
function find_difference(lab1, lab2: array of Extended) : Extended;
var
_c1,_c2,_dc,_dl,_da,_db,_dh,_first,_second,_third : Extended;
begin
_c1 := sqrt(sqr(lab1[1])+sqr(lab1[2]));
_c2 := sqrt(sqr(lab2[1])+sqr(lab2[2]));
_dc := (_c1)-(_c2);
_dl := lab1[0]-lab2[0];
_da := lab1[1]-lab2[1];
_db := lab1[2]-lab2[2];
_dh := sqrt(sqr(_da)+sqr(_db)-sqr(_dc));
_first := _dl;
_second := _dc/(1+0.045*_c1);
_third := _dh/(1+0.015*_c1);
result := sqrt(sqr(_first)+sqr(_second)+sqr(_third));
end;
reference: CIE94 - http://en.wikipedia.org/wiki/Color_difference#CIE76
|
|
|
w2m
USA
1990 Posts |
Posted - Oct 28 2012 : 17:57:32
|
I am not an expert with respect to colors.... but show us how you are adding the color values to Lab1 and Lab2. That may be the problem and not the function. When I run the function on Delphi 2010 here, I do not get any exceptions, but the result for the colors I selected returned a color difference of 0.
procedure TFormColorDifference.cxButton1Click(Sender: TObject);
var
iColorDifference: Extended;
begin
FLab1[0] := Color1.ColorValue;
FLab1[1] := Color3.ColorValue;
FLab2[2] := Color5.ColorValue;
FLab2[0] := Color2.ColorValue;
FLab2[1] := Color4.ColorValue;
FLab2[1] := Color6.ColorValue;
iColorDifference := FindColorDifference(FLab1, FLab1);
LabelColorDifference1.Caption := 'Color Difference: ' + FloatToStr(iColorDifference);
end;
William Miller
Email: w2m@frontiernet.net
EBook: http://www.imageen.com/ebook/
Apprehend: http://www.frontiernet.net/~w2m/index.html |
 |
|
|
Dudskoni
4 Posts |
Posted - Oct 29 2012 : 15:02:41
|
Dear William
Thanks for your kind assistance. This is the way I call this funcion in the main. The program works fine with the black and white part but when I add the orange and other colors the error comes up. First I know there is no need to a for loop for each color but this is something I've tried. I am no expert in Delphi so I am in trial error process now.
/////loop for black color
for i:= ImageEnView1.SelY1 to ImageEnView1.SelY2 do
for j:= ImageEnView1.SelX1 to ImageEnView1.SelX2 do
if ImageEnView1.isPointinsideSelection(j, i) then
begin
color := ImageEnView1.IEBitmap.Pixels[j,i];
lab2[0] := rgbtolab(color)[0];
lab2[1] := rgbtolab(color)[1];
lab2[2] := rgbtolab(color)[2];
//black definition
lab1[0] := 0;
lab1[1] := 0;
lab1[2] := 0;
difference := find_difference(lab1, lab2);
if (difference < 20) then
begin
inc(black);
ImageEnview1.IEBitmap.pixels_ie24RGB[j, i] := CreateRGB(0, 0, 0);
end;
end;
///// end loop color black
////loop for white color
for i:= ImageEnView1.SelY1 to ImageEnView1.SelY2 do
for j:= ImageEnView1.SelX1 to ImageEnView1.SelX2 do
if ImageEnView1.isPointinsideSelection(j, i) then
begin
color := ImageEnView1.IEBitmap.Pixels[j,i];
lab4[0] := rgbtolab(color)[0];
lab4[1] := rgbtolab(color)[1];
lab4[2] := rgbtolab(color)[2];
//White definition
lab3[0] := 100;
lab3[1] := 1;
lab3[2] := 1;
difference := find_difference(lab3, lab4);
if (difference < 20) then
begin
inc(white);
ImageEnview1.IEBitmap.pixels_ie24RGB[j, i] := CreateRGB(255, 255, 255);
end;
end;
//// end loop white color
////loop for orange color
for i:= ImageEnView1.SelY1 to ImageEnView1.SelY2 do
for j:= ImageEnView1.SelX1 to ImageEnView1.SelX2 do
if ImageEnView1.isPointinsideSelection(j, i) then
begin
color := ImageEnView1.IEBitmap.Pixels[j,i];
lab6[0]:= rgbtolab(color)[0];
lab6[1] := rgbtolab(color)[1];
lab6[2] := rgbtolab(color)[2];
//Orange definition
lab5[0] := 67;
lab5[1] := 43;
lab5[2] := 74;
difference := find_difference(lab5, lab6);
if (difference < 20) then
begin
inc(orange);
ImageEnview1.IEBitmap.pixels_ie24RGB[j, i] := CreateRGB(255, 128, 0);
end;
end;
//// end loop orange
This is the rgbtolab funcion:
function rgbtolab (RGBparaconveter: TRGB) : tdoublearray;
var
_RGBToFind_r, _RGBToFind_g, _RGBToFind_b, _x, _y, _z, _l,_a,_b : double;
begin
_RGBToFind_r := RGBparaconveter.r / 255;
_RGBToFind_g := RGBparaconveter.g / 255;
_RGBToFind_b := RGBparaconveter.b / 255;
// adjusting values
if(_RGBToFind_r > 0.04045) then
begin
_RGBToFind_r := (_RGBToFind_r+0.055)/1.055;
_RGBToFind_r := Power(_RGBToFind_r,2.4);
end
else _RGBToFind_r := _RGBToFind_r / 12.92;
if(_RGBToFind_g > 0.04045) then
begin
_RGBToFind_g := (_RGBToFind_g+0.055) / 1.055;
_RGBToFind_g := Power(_RGBToFind_g,2.4);
end
else _RGBToFind_g := _RGBToFind_g / 12.92;
if (_RGBToFind_b>0.04045) then
begin
_RGBToFind_b := (_RGBToFind_b+0.055) / 1.055;
_RGBToFind_b := Power(_RGBToFind_b,2.4);
end
else _RGBToFind_b := _RGBToFind_b / 12.92;
_RGBToFind_r := _RGBToFind_r*100;
_RGBToFind_g := _RGBToFind_g*100;
_RGBToFind_b := _RGBToFind_b*100;
_x := (_RGBToFind_r * 0.4124 + _RGBToFind_g * 0.3576 + _RGBToFind_b * 0.1805)/95.047;
_y := (_RGBToFind_r * 0.2126 + _RGBToFind_g * 0.7152 + _RGBToFind_b * 0.0722)/100;
_z := (_RGBToFind_r * 0.0193 + _RGBToFind_g * 0.1192 + _RGBToFind_b * 0.9505)/108.883;
if (_x>0.008856) then
begin
_x := power(_x,1/3);
end
else _x := (7.787*_x) + (16/116);
if (_y>0.008856) then
begin
_y := power(_y,1/3);
end
else _y := (7.787*_y) + (16/116);
if (_z>0.008856) then
begin
_z := power(_z,1/3);
end
else _z := (7.787*_z) + (16/116);
_l:= (116*_y) -16;
_a:= 500*(_x-_y);
_b:= 200*(_y-_z);
result[0] := round(_l);
result[1] := round(_a);
result[2] := round(_b);
end;
|
|
|
| |
Topic |
|
|
|
Search
Lab comparison
