how to make hsv converted image two dimensional?

I have converted an image to hsv image...now i need to find the edges by using canny.But while doing i am getting an error...

Error using ==> iptcheckinput Function EDGE expected its first input, I, to be two-dimensional.

Error in ==> edge>parse_inputs at 541 iptcheckinput(I,{'numeric','logical'},{'nonsparse','2d'},mfilename,'I',1);

Error in ==> edge at 197 [a,method,thresh,sigma,thinning,H,kx,ky] = parse_inputs(varargin{:});

Error in ==> mergealgo at 37 im_canny = edge(inputgray, 'canny');

how to make hsv image two dimensional?

[Image Analyst]

Is inputgray the Value channel? inputgray=hsvImage(:,:,3)????

Answer 1

[Jan]

You have to decide for one channel, e.g.:

RGB = rand(100, 200, 3);
HSV = rgb2hsv(RGB);
im_canny = edge(HSV(:, :, 1), 'canny');

Depending on your problem the H, S or V channel is best to recognize edges.

[vysakh]

but it now shows an error that

??? Undefined function or method '_colonobj' for input arguments of type 'double' and attributes 'full 3d real'

[Walter Roberson]

It sounds to me as if you missed a comma or two. Please show your current code.

[vysakh]

??? Undefined function or method '_colonobj' for input arguments of type 'double' and

attributes 'full 3d real'.

Error in ==> findallneighs at 8

for i=1:(max(max(watimg)))

Error in ==> combineregion at 9

neighall=findallneighs(watimg);

Error in ==> mergealgo at 48

watimg=combineregion(inputgray,watimg,originalwat,THRESH,im_canny);

when i use rgb2gray(),the value of watimg is 24..but now i am getting 5005

[Walter Roberson]

Please show size(inputgray), size(watimg), class(inputgray), class(watimg)

[vysakh]

size(inputgray)

ans =

256 256 3

>> class(watimg)

ans =

double

>> size(watimg)

ans =

256 256 3

>> class(inputgray)

ans =

double

[Walter Roberson]

a 256 x 256 x 3 image is *not* a grayscale image.

max(max(watimg)) when watimg is 3d, is going to be a vector in the third dimension. A vector cannot appear in a range specification such as 1:N . You could get around the complaint about it being "full 3d real" by using

for i = 1 : squeeze(max(max(watimg)))

but then you will still get a complaint because you will be getting a vector rather than a scalar.

If you want to be cutting edge, you could change the line to

for i = 1 : max(watimg(:))

but I think it very likely you will run into problems further on.

But is combineregion your own routine or someone else's? It is not a Mathworks-provided routine, and it does not appear to be part of any File Exchange contribution.

Answer 2

[Image Analyst]

My guess is that the "inputgray" image is still a 3D RGB image, or else the 3D HSV image. What does this say when you put it in?

[rows columns numberOfColorChannels] = size(inputgray)

It should tell you the sizes/dimensions in the command window. Here's code for a working demo using the standard MATLAB peppers color image.

clc;    % Clear the command window.
close all;  % Close all figures (except those of imtool.)
imtool close all;  % Close all imtool figures.
clear;  % Erase all existing variables.
workspace;  % Make sure the workspace panel is showing.
fontSize = 16;

% Read in a standard MATLAB color demo image.
folder = fullfile(matlabroot, '\toolbox\images\imdemos');
baseFileName = 'peppers.png';
fullFileName = fullfile(folder, baseFileName);
% Get the full filename, with path prepended.
fullFileName = fullfile(folder, baseFileName);
if ~exist(fullFileName, 'file')
  % Didn't find it there.  Check the search path for it.
  fullFileName = baseFileName; % No path this time.
  if ~exist(fullFileName, 'file')
    % Still didn't find it.  Alert user.
    errorMessage = sprintf('Error: %s does not exist.', fullFileName);
    uiwait(warndlg(errorMessage));
    return;
  end
end
rgbImage = imread(fullFileName);
% Get the dimensions of the image.  numberOfColorBands should be = 3.
[rows columns numberOfColorBands] = size(rgbImage);
% Display the original color image.
subplot(2, 4, 1);
imshow(rgbImage, []);
title('Original Color Image', 'FontSize', fontSize);
% Enlarge figure to full screen.
set(gcf, 'units','normalized','outerposition',[0 0 1 1]);

% Get HSV color space version of the image
hsv = rgb2hsv(rgbImage);
h = hsv(:, :, 1);
s = hsv(:, :, 2);
v = hsv(:, :, 3);
% Display the images.
subplot(2, 4, 2);
imshow(h, []);
title('Hue Image', 'FontSize', fontSize);
subplot(2, 4, 3);
imshow(s, []);
title('Saturation Image', 'FontSize', fontSize);% Display the images.
subplot(2, 4, 4);
imshow(v, []);
title('Value Image', 'FontSize', fontSize);

% Run Canny Edge on them all
hEdge = edge(h, 'canny');
sEdge = edge(s, 'canny');
vEdge = edge(v, 'canny');

% Display the images.
subplot(2, 4, 6);
imshow(hEdge, []);
title('Hue Edge Image', 'FontSize', fontSize);
subplot(2, 4, 7);
imshow(sEdge, []);
title('Saturation Edge Image', 'FontSize', fontSize);% Display the images.
subplot(2, 4, 8);
imshow(vEdge, []);
title('Value Edge Image', 'FontSize', fontSize);

[vysakh]

i want the same image to be converted to hsv and use it to find edges using canny...

[Image Analyst]

And what do you think I did? Did you even run this demo I made for you???

Answer 3

[Walter Roberson]

There is no 2 dimensional equivalent to an hsv image.

The closest you can get is to use the hsv equivalent of pseudo-color images -- to create a 2 dimensional array in which each value is the index of an HSV tuple in a map.

You can create such a thing by applying rgb2ind() to the HSV array (rgb2ind() will not notice that the data is actually HSV.)

I warn, however, that the canny edge detector will likely have very poor performance on mapped images. The index numbers that are used in mapping do not have any relationship to each other: index #83 might refer to a color that is extremely different than index #84.

If working on one of the individual channels, H, S, or V, does not work for you, then converting to a mapped HSV image and doing the edge detection on that will almost certainly be much worse.

But you can always try it if you need to prove to yourself that a "2 dimensional HSV image" is the wrong thing to use.