Svm+surf+k-means flower classification (matlab)

Detailed algorithm 、 For the steps, see python edition ：
https://editor.csdn.net/md/?articleId=103732613

Training documents

clc
clear all
train_path = "g:/flowers/train/";
test_path = "g:/flowers/test/";
K = 100;
build_centers(train_path,K);
% Construct training set feature vector 
[x_train,y_train] = cal_vec(train_path,K);
% Input eigenvectors and labels into SVM In the classifier 
Train(x_train,y_train);
% Calculate the accuracy of the test set 
[predicted_label, accuracy, decision_values] = Test(test_path,K);

function f1 = calcSurfFeature(img)
% Convert the image to a grayscale image 
gray = rgb2gray(img);
% Calculation surf Characteristic point 
points = detectSURFFeatures(gray);
% Calculate the description vector 
[f1, vpts1] = extractFeatures(gray, points);
end

%% Calculation word bag 
function centers = learnVocabulary(features,K)
% Set the number of word bags 
% use k-means to cluster a bag of features
%[ Cluster number , Location of points ] = [ Input , Number of clusters , The initial center point , Number of cluster repetitions ]
opts = statset('Display','final','MaxIter',1000);
[idx ,centers] = kmeans(features ,K,'Replicates',10, 'Options',opts);
end

%% Calculate the feature vector of the picture according to the dictionary 
function featVec = calcFeatVec(features, centers,K)
featVec = zeros(1,K);
%m Represents the number of characteristic points ,n Dimension representing feature points 
[m,n] = size(features);
for i = 1:m
    fi = features(i,:);
    %% Is characterized by 1*64, And the dictionary is K*64, We need to find out which word bag this feature is closest to 
    res = zeros(K,64);
    for j = 1:K
        res(j,:) = fi;
    end
    % Copy the feature vertically into K*64
    % Find the distance 
    distance = sum((res - centers).^2,2);
    distance = sqrt(distance);
    % Sort the distances , Find the minimum distance 
    [x,y] = min(distance);
    featVec(y) = featVec(y)+1;
end
end

%%% Build a dictionary 
function build_centers(path,K)
% Get the path subdirectory , Save the subdirectory path to cate in 
sub = dir(path);
cate = [];
features = [];
for i = 1:size(sub)
    if sub(i).name ~= "." && sub(i).name ~= ".." && sub(i).isdir
        s = strcat(strcat(path ,sub(i).name),"/");
        cate = [cate s];
    end
end
for i = 1:length(cate)
    imgDir  = dir( strcat(cate(i),'*.jpg'));
    for j = 1:length(imgDir)
        img = imread(strcat(cate(i),imgDir(j).name));
        % Calculation surf features 
        img_f = calcSurfFeature(img);
        % Feature points are added to the feature set 
        features = [features;img_f];
    end
end
[m,n] = size(features);
fprintf(" Training set feature point set :[ %d , %d ]\n",m,n);
centers = learnVocabulary(features,K);
filename = "g:/flowers/svm/svm_centers.mat";
save(filename,'centers');
end

%% according to centers Calculate the image feature vector 
function [data_vec,labels] = cal_vec(path,K)
load("g:/flowers/svm/svm_centers.mat");
% Store image feature vectors 
data_vec = [];
% Store picture labels （ Category ）
labels = [];
% Get the subdirectories under the current directory 
sub = dir(path);
cate = [];
for i = 1:size(sub)
    if sub(i).name ~= "." && sub(i).name ~= ".." && sub(i).isdir
        s = strcat(strcat(path ,sub(i).name),"/");
        cate = [cate s];
    end
end
for i = 1:length(cate)
    imgDir  = dir( strcat(cate(i),'*.jpg'));
    for j = 1:length(imgDir)
        img = imread(strcat(cate(i),imgDir(j).name));
        % Calculation surf features 
        img_f = calcSurfFeature(img);
        img_vec = calcFeatVec(img_f, centers,K);
        data_vec = [data_vec;img_vec];
        labels = [labels,i];
    end
end
end

% Training SVM classifier 
function Train(data_vec,labels)
modle = libsvm_train(labels',data_vec,'-t 0');
save("g:/flowers/svm/svm_model.mat",'modle');
fprintf("Training Done!\n");
end

%SVM Test the correctness of the test set 
function [predicted_label, accuracy, decision_values] = Test(path,K)
% Read SVM Model 
load("g:/flowers/svm/svm_model.mat");
% Read dictionary 
load("g:/flowers/svm/svm_centers.mat");
% Calculate the eigenvector of each picture 
[data_vec,labels] = cal_vec(path,K);
[num_test , y] = size(data_vec);
[predicted_label, accuracy, decision_values] = libsvm_predict(labels',data_vec,modle);
fprintf("Testing Done!\n");
end

Forecast file

clc
clear all
img_path = "e:\minist\predict\";
class_flower = ["0","1","2","3","4","5","6","7","8","9"];
K = 100;
for i = 1:5
    path = strcat(strcat(img_path,num2str(i)),".jpg");
    img = imread(path);
    figure(i);
    imshow(img);
    res = predict(img,K);
    xlabel([" Predicted results ：",class_flower(res)]);
end

% Predict a single picture 
function res = predict(img,K)
load("g:/flowers/svm/svm_centers.mat");
load("g:/flowers/svm/svm_model.mat");
features = calcSurfFeature(img);
featVec = calcFeatVec(features, centers,K);
res = libsvm_predict([1],featVec,modle);
end

function f1 = calcSurfFeature(img)
% Convert the image to a grayscale image 
gray = rgb2gray(img);
% Calculation suft Characteristic point 
points = detectSURFFeatures(gray);
% Calculate the description vector 
[f1, vpts1] = extractFeatures(gray, points);
end

%% Calculate the feature vector of the picture according to the dictionary 
function featVec = calcFeatVec(features, centers,K)
featVec = zeros(1,K);
%m Represents the number of characteristic points ,n Dimension representing feature points 
[m,n] = size(features);
for i = 1:m
    fi = features(i,:);
    %% Is characterized by 1*64, And the dictionary is 50*64, We need to find out which word bag this feature is closest to 
    res = zeros(K,64);
    for j = 1:K
        res(j,:) = fi;
    end
    % Copy the feature vertically into K*64
    % Find the distance 
    distance = sum((res - centers).^2,2);
    distance = sqrt(distance);
    % Sort the distances , Find the minimum distance 
    [x,y] = min(distance);
    featVec(y) = featVec(y)+1;
end
end