How to convert the RGB color map into an HSV color map

Hi All,

I need to do these steps :

1-Convert an RGB image to colormap say M

2-Convert this RGB colormap M to an HSV colormap say cmap

By using this conversion in MATLAB cmap = rgb2hsv (M) that converts an RGB colormap M to an HSV colormap,But my question How can I get the RGB colormap M firstly then to us it in rgb2hsv(M) function? Please could anyone explain to me in example,Thanks in advance for Help.

I'll wait your answer impatiently,

[hamed abdulaziz]

I tried this(is it correct?) :

RGB=imread('d:\168.bmp');

n=65536;

% Converts the RGB image to an indexed image X ;n = colors

[IND,map] = rgb2ind(RGB,n);

% Convert RGB colormap to HSV colormap(is it correct?)

cmap=rgb2hsv(map);

% Calcula the color variance of the HSV colormap

color_variance=var(cmap);

Is the color_variance represent the actual color variance for an image,because I have three coloumns!!!

Answer 1

[Image Analyst]

No that does not make any sense. You can't just apply some different colormap to an indexed image and expect it to look sensible. It will look like garbage. And calculating the variance of a colormap is meaningless. Perhaps you want the color frequency image like this calculates but I don't know.

I think perhaps you're getting color map confused with colorspace. You want to do something, though I don't know what it is. And you think that this colormap approach is the way to do it, but it's not - it's a dead end. So why don't you tell me what you really want to do and I can recommend an approach. For example you want so segment out pixels of some particular color out of an image, or you want to classify the image into 5 different color classes.

[hamed abdulaziz]

Image Analyst, Thanks for your advice always,I read this paper:"Image Processing for Skin Cancer Features Extraction",that says :

"Color Variation: To calculate color variance, first step is to convert the RGB color map into an HSV color map. The column of the output matrix represents hue, saturation, and value respectively. In second step, the color variance of the HSV color map has been calculated which represent the actual color variance for an image. In this way, the color variance of all the skin lesions has been calculated. From the color variance data, one can draw a conclusion that the color variance of all the melanoma skin lesions is varying within the range of 0.10-0.15. ",By the how can I attach file to my comment?

[hamed abdulaziz]

Image Processing for Skin Cancer Features Extraction.pdf

[Image Analyst]

They mean hsv color SPACE, not colormap. At least using MATLAB terminology where colormap is a pseudocolor look up table that converts a gray scale value into an RGB color.

From the sounds of it, it sounds like they're converting to hsv color space with rgb2hsv. Then they find the mean hsv of the region of interest. Then they find delta E, which is the color difference, for every pixel - the delta E between that pixel and the mean color. You can do this in LAB color space also - the delta E formula in LAB color space is simpler than in HSC (LCH) color space. (See easyRGB.com) If the variance of the delta E's is high they say it's a melanoma (though I'm not sure I agree with that since melanomas can be dark and uniform, more so than the surrounding normal skin). I might go along with that for basal or squamous skin cancer more than for melanoma, but I'm not a dermatologist, just a color scientist.

You can use the image or paperclip icon above your commetn box to insert images or attach files.

[hamed abdulaziz]

Image Analyst, Could you explain your answer by code? To let me know my errors

[Image Analyst]

Try this demo: http://www.mathworks.com/matlabcentral/fileexchange/31118-color-segmentation-by-delta-e-color-difference

[hamed abdulaziz]

I did this RGB_image=imread(d:\1.bmp);

HSV_image=rgb2hsv(RGB_image)

Now how can calculat the variance of HSV_image ? will I get one value or three values?please could you complete the above code?

[Image Analyst]

What does the paper you're following say? I might do it a different way than they do it. You should get only 1 variance but you need to construct a color difference image first. You can get the variance of each h, s, and v separately but that's not the color difference between a standard color (the mean color of the lesion) and the individual pixels. An overall color difference, delta E, is comprised of changes in h, s, and v, so you can't just get variances in each or you're not getting the variance in the total color difference, you're just getting the variance in one component of the total color difference.

[hamed abdulaziz]

How can I construct a color difference image?

[Image Analyst]

Transform into lab color space with makecform(). Find the mean L, a, and b in the region of interest. Then calculate delta images:

deltaLimage = Limage - meanL;
deltaAimage = Aimage - meanA;
deltaBimage = Bimage - meanB;
% Calculate deltaE image
deltaE = sqrt(deltaLimage  .^ 2 + deltaAimage  .^ 2 + deltaBimage  .^ 2);

I have a delta E demo in my File Exchange.