Math Forum Discussions

john john

Posts: 18
Registered: 6/22/11

can someone help me to fix this error(Error using ==> kmeans at 382 An empty cluster error occurred in every replicate)
Posted: Feb 15, 2012 10:42 AM

Plain Text

??? Error using ==> kmeans at 382 An empty cluster error occurred in every replicate.

Error in ==> segmentizeImage1 at 25 [cluster_idx cluster_center] = kmeans(ab,nColors,'distance','sqEuclidean', ...

Error in ==> bui_simple>start_snap_Callback at 108 segmentizeImage1;

Error in ==> gui_mainfcn at 96 feval(varargin{:});

Error in ==> bui_simple at 42 gui_mainfcn(gui_State, varargin{:});

Error in ==> @(hObject,eventdata)bui_simple('start_snap_Callback',hObject,eventdata,guidata(hObject))

??? Error while evaluating uicontrol Callback

the code is working, it can captures and recognize data, the problem later on it will goes into that error, can some1 explain it to me why it give's error? is it bug on kmeans.m on toolbox?

REFER TO THIS CODE BELOw:
x = wa; % this is where the image get from snapshot on video
rawImage = x;
I = imresize(rawImage,0.50);
cform = makecform('srgb2lab');
J = applycform(I,cform);
ab = double(J(:,:,2:3));
nrows = size(ab,1);
ncols = size(ab,2);
ab = reshape(ab,nrows*ncols,2);
nColors = 2;
[cluster_idx cluster_center] = kmeans(ab,nColors,'distance','sqEuclidean', ...
'Replicates',2);
pixel_labels = reshape(cluster_idx,nrows,ncols);
segmented_images = cell(1,2);
rgb_label = repmat(pixel_labels,[1 1 3]);
for k = 1:nColors
color = I;
color(rgb_label ~= k) = 0;
segmented_images{k} = color;
end
mean_cluster_value = mean(cluster_center,2);
[tmp, idx] = sort(mean_cluster_value);
hand_cluster_num = idx(2);
K = segmented_images{hand_cluster_num};
L = im2bw(segmented_images{hand_cluster_num},0.001);
s = regionprops(rgb_label, I, {'Centroid','BoundingBox'});
% s(hand_cluster_num).Centroid
% s(hand_cluster_num).BoundingBox
centroid_x = uint8(s(hand_cluster_num).Centroid(1));
centroid_y = uint8(s(hand_cluster_num).Centroid(2));
cropped_x = uint8(s(hand_cluster_num).BoundingBox(1))+1;
cropped_y = uint8(s(hand_cluster_num).BoundingBox(2))+1;
cropped_w = uint8(s(hand_cluster_num).BoundingBox(4))-2;
cropped_h = uint8(s(hand_cluster_num).BoundingBox(5))-2;
% apply gaussian blur to remove noise
PSF = fspecial('gaussian',7,10);
L = imfilter(L,PSF,'symmetric','conv');
% cropped region of interest from main image
M = L( cropped_y : (cropped_y+cropped_h), cropped_x: (cropped_x+cropped_w) );
% trim leading and leading zeroes
M(:, logical(sum(abs(M)) == 0)) = [];
cropped_h = size(M,1);
cropped_w = size(M,2);
non_zeroes = find(sum(abs(L))~=0);
cropped_x = non_zeroes(1);
% fill holes to remove flares
M = imfill(M, 'holes');
% get a shrunk sample from cropped image
N = imresize(M, 0.25);
[sample_height sample_width] = size(N);
% clip excess pixels
if(sample_height>50)
N = N(1:50,:);
end
if(sample_width>50)
N = N(:,1:50);
end
[sample_height sample_width] = size(N);
%pad both sides
htopad = double(25 - ceil(sample_height/2));
wtopad = double(25 - ceil(sample_width/2));
if((htopad>=0)&&(wtopad>=0))
N = padarray(N,[htopad wtopad]);
%for uneven padding, add padding to end
[sample_height sample_width] = size(N);
htopad = 50 - sample_height;
wtopad = 50 - sample_width;
N = padarray(N,[htopad wtopad],0,'post');
else
disp('Padding not performed...');
end
imshow(M);
%hold on;
% plot(centroid_x, centroid_y, 'bo');
% rectangle('Position', [ cropped_x cropped_y cropped_w cropped_h ], 'EdgeColor','y');
% hold off;
y = N;
z = imresize(y, [40 30]);
z = z(:);