How to extract RGB

I'm supposed to "extract rgb" values from 5 points from an image and I am very confused on how to do it. I am then supposed to get each of those RGB and convert it to HSI so that I am supposed to have 5 HSI. It sounds like I have to create 5 rgb variables, right? If so, what is the function I need to use?

[Geoff Hayes]

Davidson - is your image three dimensional (and so has the red, green, and blue components)? Are you unsure how to extract the red component for one of these five points?

[Davidson Bock]

its a simple 2d image. Here are part of the instructions ask

 Get 5 user clicks. Use ginput() function.
 Make sure to display image before calling ginput() to see the image to click on.
 Extract RGB value for each user click
 Convert each RGB to HSI.  → so you should have 5 hue values

I already did the 1st 2 parts of this piece of the instructions. This is my code so far.

 f = imread ('peppers_preview.jpeg.jpg');
 imshow(f)
 [x,y] = ginput (5);
 for i = 1:5
 red = f(int16(y(i)), int16(x(i)), 1)
 green = f(int16(y(i)), int16(x(i)), 2)
 blue = f(int16(y(i)), int16(x(i)), 3)
end

Answer 1

[RobF]

Extraction of RGB values:

C = imread('ngc6543a.jpg'); % Sample image file
% Coordinates of point/pixel P1: x1, y1
x1 = 354;   % Sample coordinates
y1 = 284;   % Sample coordinates
% Extract values for red (R), green (G) and blue (B):
R1 = C(x1, y1, 1)   % Red is first layer
G1 = C(x1, y1, 2)   % Green is second layer
B1 = C(x1, y1, 1)   % Blue is third layer

Information on RGB to HSI conversion:

https://www.mathworks.com/matlabcentral/fileexchange/53354-hsi-%253C-%253E-rgb-image-conversion-tools?s_tid=answers_rc2-1_p4_MLT

[Davidson Bock]

Oh, I think I already did that. I have the code of what I did in the comments.

Answer 2

[DGM]

I'm just going to throw this down here. MATLAB has no native HSI conversion tools. It has HSV tools, but if you actually want HSI, you'll have to find third-party conversion tools. @RobF linked to the tools from MIMT. I'll continue on the assumption that MIMT is available.

This will collect up to 5 points (discarding any invalid points). Results will be returned as color tables in RGB and HSI as instructed. For sake of comparison, I'm also converting a copy to LAB as well.

% load an image
inpict = imread('peppers.png');

% input must be RGB
% MIMT imsize() will return fixed-length vector output
% so we can know that sz(3) will always exist, and that it will be 
% correct even if there are trailing non-singleton dimensions
sz = imsize(inpict,3); % size of first three dims
if sz(3)~=3
    error('input must be RGB')
end

% display it and collect 5 points
imshow(inpict);
[x y] = ginput(5);

% round to nearest pixel
x = round(x);
y = round(y);

% discard points which lie outside image area
badpts = x<1 | x>sz(2) | y<1 | y>sz(1);

% generate index list
x = repmat(x(~badpts),[1 sz(3)]);
y = repmat(y(~badpts),[1 sz(3)]);
ch = repmat(1:sz(3),[size(x,1) 1]);
idx = sub2ind(sz,y,x,ch);

% selected points as a Px3 color table of unit-scale RGB tuples
CTrgb = im2double(inpict(idx))

CTrgb =

0.2627 0.1451 0.2706

0.2706 0.1608 0.2902

0.2784 0.1569 0.2902

0.3020 0.1686 0.2902

0.2745 0.1608 0.2863

% selected points as a Px3 color table of LAB tuples
% native conversion tools accept color tuples/tables directly
CTlab = rgb2lab(CTrgb)

CTlab =

19.8274 20.2164 -14.2700

21.3782 19.5417 -15.2412

21.4010 20.9113 -15.1841

22.9996 21.2279 -12.6460

21.4656 19.6757 -14.4189

% selected points as a Px3 color table of HSI tuples
% MIMT conversion tools generally don't accept color tables due to the ambiguity
% so permute the CT so that it can be processed explicitly as a Px1x3 image
% MIMT ctflop(mat) is the same as permute(mat,[1 3 2])
% this is an involution, so the second call returns the array to Px3
CThsi = ctflop(rgb2hsi(ctflop(CTrgb)))

CThsi =

296.8021 0.3584 0.2261

291.9196 0.3315 0.2405

295.4294 0.3514 0.2418

304.5706 0.3351 0.2536

295.1311 0.3315 0.2405