Using imwrite() to create a .bmp file with custom colormap

I am trying to use the imwrite() function to create and save a .bmp file using a custom colormap, 'myMap'. The colormap contains values between 0 and 1 and is a variable of type double with size [51 3]. My image matrix is of type double and contains values 0 to 1. When I produce this image with imagesc() it comes out fine:

figure
imagesc(myImage)
axis equal
axis tight
colormap(myMap)

However, when I use imwrite to save it as a .bmp file the image is saved in all one dark grey color:

imwrite(myImage, myMap, 'grid.bmp')

I have tried converting myImage to uint8 format and scaling myMap into [256 3] size, but neither works. What am I doing wrong? Thanks.

Answer 1

[Walter Roberson]

The colormap for the imwrite must be in the range 0 to 1.

When you imwrite with a colormap the data must be integral. If it is type double or single then 1 is subtracted from it and the result is converted to uint8. In other words the values must already be colormap indices.

You need:

Numcolor = size(myMap, 1);
Imgmin = min(MyImage(:)) ;
Imgmax = max(MyImage(:)) ;
mappedImage = uint8( (MyImage-Imgmin)./(Imgmax-ImgMin).* (Numcolor-1) ) ;

Then imwrite(mappedImage, myMap,...)

[Hailey Trier]

Great, this solved my problem! Thanks.

Answer 2

[Image Analyst]

You can convert your image to RGB with ind2rgb():

rgbImage = ind2rgb(myImage, myMap);
imwrite(rgbImage, filename);

[Hailey Trier]

Using this solution I get the error:

Error using dither>parse_inputs (line 92)
DITHER(RGB,map): the RGB image has to be a three-dimensional array.

Error in dither (line 47)
[X,m,qm,qe] = parse_inputs(varargin{:});

Error in rgb2ind (line 106)
        X = dither(RGB,map);

[Image Analyst]

Try ind2rgb() instead of rgb2ind().

[Walter Roberson]

That will not work as the user's original data is not indices, it is data with an arbitrary range. Notice they used imagesc() to display it.

[Image Analyst]

Right, you have to normalize like you did. This will work:

myImage = im2double(imread('cameraman.tif'));
% myImage is floating point in the range 0-1
myMap = hsv(256); % Whatever one you want.
% Convert to gray levels (indexes):
myImage = uint8(255 * myImage);
% Now convert to RGB color.
rgbImage = ind2rgb(myImage, myMap);
imshow(rgbImage);

If you want to scale the min to 0 and max to 255, you can use mat2gray():

myImage = uint8(255 * mat2gray(myImage));

[Walter Roberson]

They are working with data not an image so they need to normalize to 0 to 1 and then to integer indices like I coded

[Image Analyst]

Why do you say it's non-image data when she called the variable "myImage", used imagesc() to display it, and tried to save it with imwrite()? As far as I can guess, it's a floating point (double) image. Anyway, my code also normalizes and converts to integer indices like yours, just using an alternate method.

[Dayangku Nur Faizah Pengiran Mohamad]

Hi. I want ask a question.

How do I save the original image jpg from figure for colormap?

Here my codes:

>> I=imread('M_92_ar_2001743_st01_currentsize_200x200.jpg');

>> imshow(I);

>> imfinfo('M_92_ar_2001743_st01_currentsize_200x200.jpg');

>> I2=I(:,:,1);

>> imshow(I2);

>> map = [0, 0.3, 0

0, 0.4, 0

0, 0.5, 0

0, 0.6, 0

0, 0.8, 0

0, 1.0, 0];

>> imshow(I2);

>> colormap(map);

>> imwrite(I2,map,'cmalignant200.jpg');