i want to classify color feature of dataset with SVM, it can or not?

Question

Anggita Puspawardani 2019-5-19

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/462893-i-want-to-classify-color-feature-of-dataset-with-svm-it-can-or-not

评论： Anggita Puspawardani 2020-10-12

Hi, I'm a newbee of Matlab. I have a problem that i want to classify with SVM, but i cannot. For segmentation, I'm using K-Means. And for feature extraction, I'm using Histogram.

I just want to classify of dataset based on color feature. So, what should i do?

here's my code:

clc;clear;close all;
[filename,pathname] = uigetfile({'*.*';'*.bmp';'*.tif';'*.gif';'*.png'},'Pick a Disease Affected Leaf');
I = imread([pathname,filename]);
figure, imshow(I);title('Disease Affected Leaf');
% Color Image Segmentation Using K-Means Clustering
cform = makecform('srgb2lab');
lab_he = applycform(I,cform);
figure, imshow(lab_he), title('L*a*b color space');
ab = double(lab_he(:,:,2:3));
nrows = size(ab,1);
ncols = size(ab,2);
ab = reshape(ab,nrows*ncols,2);
nColors = 3;
[cluster_idx cluster_center] = kmeans(ab,nColors,'distance','sqEuclidean', ...
                                      'Replicates',3);
pixel_labels = reshape(cluster_idx,nrows,ncols);
RGB = label2rgb(pixel_labels);
figure, imshow(RGB,[]), title('image labeled by cluster index');
segmented_images = cell(1,3);
rgb_label = repmat(pixel_labels,[1,1,3]);
for k = 1:nColors
    colors = I;
    colors(rgb_label ~= k) = 0;
    segmented_images{k} = colors;
end
figure, subplot(3,1,1);imshow(segmented_images{1});title('Cluster 1'); subplot(3,1,2);imshow(segmented_images{2});title('Cluster 2');
subplot(3,1,3);imshow(segmented_images{3});title('Cluster 3');
% Leaf segmentation
area_cluster1 = sum(sum(pixel_labels==1));
area_cluster2 = sum(sum(pixel_labels==2));
area_cluster3 = sum(sum(pixel_labels==3));
[~,cluster_leaf] = min([area_cluster1,area_cluster2,area_cluster3]);
leaf_bw = (pixel_labels==cluster_leaf);
leaf_bw = imfill(leaf_bw,'holes');
leaf_bw = bwareaopen(leaf_bw,1000);
leaf = I;
R = leaf(:,:,1);
G = leaf(:,:,2);
B = leaf(:,:,3);
R(~leaf_bw) = 0;
G(~leaf_bw) = 0;
B(~leaf_bw) = 0;
leaf_rgb = cat(3,R,G,B);
figure, imshow(leaf_rgb), title('the leaf only (RGB Color Space)');
% RGB Features Extraction
R_stats = regionprops(leaf_bw,R,'PixelValues','MeanIntensity',...
    'MaxIntensity','MinIntensity');
G_stats = regionprops(leaf_bw,G,'PixelValues','MeanIntensity',...
    'MaxIntensity','MinIntensity');
B_stats = regionprops(leaf_bw,B,'PixelValues','MeanIntensity',...
    'MaxIntensity','MinIntensity');
 
R_pix_val = R_stats.PixelValues;
G_pix_val = G_stats.PixelValues;
B_pix_val = B_stats.PixelValues;
 
figure,
histogram(R_pix_val,256,'FaceColor','r','EdgeColor','r')
set(gca,'XLim',[0 255])
set(gca,'YLim',[0 15000])
grid on
title('Histogram of Red Channel')
 
figure,
histogram(G_pix_val,256,'FaceColor','g','EdgeColor','g')
set(gca,'XLim',[0 255])
set(gca,'YLim',[0 15000])
grid on
title('Histogram of Green Channel')
 
figure,
histogram(B_pix_val,256,'FaceColor','b','EdgeColor','b')
set(gca,'XLim',[0 255])
set(gca,'YLim',[0 15000])
grid on
title('Histogram of Blue Channel')
R_mean = R_stats.MeanIntensity;
G_mean = G_stats.MeanIntensity;
B_mean = B_stats.MeanIntensity;
 
R_max = R_stats.MaxIntensity;
G_max = G_stats.MaxIntensity;
B_max = B_stats.MaxIntensity;
 
R_min = R_stats.MinIntensity;
G_min = G_stats.MinIntensity;
B_min = B_stats.MinIntensity;
 
% HSV features extraction
leaf = rgb2hsv(I);
H = leaf(:,:,1);
S = leaf(:,:,2);
V = leaf(:,:,3);
H(~leaf_bw) = 0;
S(~leaf_bw) = 0;
V(~leaf_bw) = 0;
leaf_hsv = cat(3,H,S,V);
figure, imshow(leaf_hsv), title('the leaf only (HSV Color Space)');
 
H_stats = regionprops(leaf_bw,H,'PixelValues','MeanIntensity',...
    'MaxIntensity','MinIntensity');
S_stats = regionprops(leaf_bw,S,'PixelValues','MeanIntensity',...
    'MaxIntensity','MinIntensity');
V_stats = regionprops(leaf_bw,V,'PixelValues','MeanIntensity',...
    'MaxIntensity','MinIntensity');
 
H_mean = H_stats.MeanIntensity;
S_mean = S_stats.MeanIntensity;
V_mean = V_stats.MeanIntensity;
 
H_max = H_stats.MaxIntensity;
S_max = S_stats.MaxIntensity;
V_max = V_stats.MaxIntensity;
 
H_min = H_stats.MinIntensity;
S_min = S_stats.MinIntensity;
V_min = V_stats.MinIntensity;
% Print the features in the command window.
disp('>>> Features Extracted <<<')
disp('<1> Red Channel')
fprintf('Mean = %6.2f \n',R_mean);
fprintf('Max  = %6.0f \n',R_max);
fprintf('Min  = %6.0f \n',R_min);
 
disp('<2> Green Channel')
fprintf('Mean = %6.2f \n',G_mean);
fprintf('Max  = %6.0f \n',G_max);
fprintf('Min  = %6.0f \n',G_min);
 
disp('<3> Blue Channel')
fprintf('Mean = %6.2f \n',B_mean);
fprintf('Max  = %6.0f \n',B_max);
fprintf('Min  = %6.0f \n',B_min);
 
disp('<4> Hue Channel')
fprintf('Mean = %6.4f \n',H_mean);
fprintf('Max  = %6.4f \n',H_max);
fprintf('Min  = %6.4f \n',H_min);
 
disp('<5> Saturation Channel')
fprintf('Mean = %6.4f \n',S_mean);
fprintf('Max  = %6.4f \n',S_max);
fprintf('Min  = %6.4f \n',S_min);
 
disp('<6> Value Channel')
fprintf('Mean = %6.4f \n',V_mean);
fprintf('Max  = %6.4f \n',V_max);
fprintf('Min  = %6.4f \n',V_min);
%% Classification Using SVM
% Load the training set
load Diseaseset.mat
% 'diasesefeat' contains the features of the disease affected leaves of both
% the types
% 'diseasetype' contains the corresponding label
% Train the classifier 
svmStructDisease = fitcsvm(diseasefeat,diseasetype);
% Classify the test leaf
species_disease = predict(svmStructDisease,feat_disease)
% Observe the results on the command window