%% 图片数据库转化为特征向量
clc
clear


% 本地folder 图片构建特征向量
errorDownload_file = 'D:\Program Files\MATLAB\R2018a\work\特征提取\outputData\errorDownload_file.txt';

DataSetFolder = 'D:\Program Files\MATLAB\R2018a\work\特征提取\dataset\home textiles-v1.0'; %  数据文件夹路径
% DataSetFolder = 'E:\Program Files\MATLAB\R2018a\work\特征提取\dataset\home textiles-v2.0';

DataSet_outputFolder  = 'D:\Program Files\MATLAB\R2018a\work\特征提取\outputData';

DataSet_FileName = '\DataSet_FeatureVec_file.txt';

fun_ImageFileTransformDataSet(DataSetFolder, DataSet_outputFolder, DataSet_FileName, errorDownload_file);


% URLsql图片构建特征向量
errorDownload_file = 'D:\Program Files\MATLAB\R2018a\work\特征提取\outputData\errorDownload_file.txt';

DataSet_outputFolder  = 'D:\Program Files\MATLAB\R2018a\work\特征提取\outputData';

DataSet_FileName = '\DataSet_FeatureVec.txt';

DataSet_savefile = [DataSet_outputFolder, DataSet_FileName];

fun_ImageUrlSqlTransformDataSet(DataSet_outputFolder, DataSet_FileName, errorDownload_file);

%% 错误图片再尝试
fun_ImageErrorTransformDataSet(errorDownload_file, DataSet_outputFolder, DataSet_FileName)



%% 得到距离和相似度的最优转化参数
queryImage = {'http://www.591wx.com/Upload/webAdv/2017-09-20/cc97c7deb84840a19f35df27eb9b5c48.jpg'};
Parameter_Dist2Similarity_best = fun_getBestParmForSimilarity(queryImage, DataSet_outputFolder, DataSet_FileName);


%% 聚类
% todo:使用tall得到聚类中心，使用mapreduce得到各个类别的样本集（非常好的构思）。

outputname = 'first';
% Cluster_outputFolder = 'D:\Program Files\MATLAB\R2018a\work\特征提取\outputData\Cluster_tall'; 
% fun_FeatureCluster(DataSet_outputFolder, DataSet_FileName, Cluster_outputFolder, outputname); % tall并行聚类，但是对60万数据集遍历了600次
Cluster_outputFolder = 'D:\Program Files\MATLAB\R2018a\work\特征提取\outputData\Cluster_parallel';
fun_FeatureCluster_parallel(DataSet_outputFolder, DataSet_FileName, Cluster_outputFolder, outputname)% mapreduce 聚类
Cluster_outputFolder = 'D:\Program Files\MATLAB\R2018a\work\特征提取\outputData\Cluster_tallMapreduce';
fun_FeatureCluster_tallMapreduce(DataSet_outputFolder, DataSet_FileName, Cluster_outputFolder, outputname)


%% 目标图像
ObjSetFolder = 'D:\Program Files\MATLAB\R2018a\work\特征提取\dataSet';
ObjSetimds = imageDatastore(ObjSetFolder);
ObjSetImageFile = ObjSetimds.Files;
[~,ObjSetImageFileName] = cellfun(@fileparts,ObjSetImageFile,'uniformOutput',false);
warning off
for i=1:length(ObjSetImageFile)
    queryImage = ObjSetImageFile{i};

    [retrievalImageHomePage,retrievalImageSimilarity] = fun_ImageRetrieval_Cluster(queryImage, Cluster_outputFolder, outputname); % todo :返回距离和相似度
    % [retrievalImageHomePage,retrievalImageSimilarity, retrievalImageURL] = fun_ImageRetrieval_NoCluster(ImageURL);
    figure
    montage(retrievalImageHomePage, 'Size', [10 10]);
    saveas(gcf,['D:\Program Files\MATLAB\R2018a\work\特征提取\fig\', ObjSetImageFileName{i},'.jpg']);

end

warning on

% % 展示目标图片i的检索结果
% subplot(3, 5, [1:5])
% image(imread(char(queryImage)));
% axis equal
% axis off
% title('目标图像')
% for j = 1:10
%     retrieval_ip = retrievalImageHomePage{j};
%     picture_data = imread(retrieval_ip);
%     subplot(3, 5, 5+j)
%     image(picture_data);
%     axis equal
%     axis off
% end


if ~exist('D:\Program Files\MATLAB\R2018a\work\特征提取\fig', 'dir') 
    mkdir(['D:\Program Files\MATLAB\R2018a\work\特征提取\fig', '\fig'])
end 
% saveas(gcf, ['D:\Program Files\MATLAB\R2018a\work\特征提取\fig',]);



%% 添加新数据
NewDataSet_outputFolder  = 'D:\Program Files\MATLAB\R2018a\work\特征提取\outputData'; % 
NewDataSet_FileName = '\NewDataSet_FeatureVec.txt'; % 新数据集文件（FeatureVec）
DataSet_outputFolder  = 'D:\Program Files\MATLAB\R2018a\work\特征提取\outputData';
DataSet_FileName = '\DataSet_FeatureVec.txt';               % 总数据集文件
errorDownload_file = 'D:\Program Files\MATLAB\R2018a\work\特征提取\outputData\errorDownload_file.txt';

fun_ImageUrlSqlTransformDataSet(DataSet_outputFolder, DataSet_FileName, NewDataSet_outputFolder, NewDataSet_FileName, errorDownload_file)

try
    load([DataSet_outputFolder, '\NumAccumulaNewDataSet.mat'])
catch
    n_AccumulaNewDataSet = 0;
end

Cluster_outputFolder = 'D:\Program Files\MATLAB\R2018a\work\特征提取\outputData\Cluster_parallel';
outputname = 'first';

n_AccumulaNewDataSet = fun_AddNewData(n_AccumulaNewDataSet,...
    NewDataSet_outputFolder,  NewDataSet_FileName,...
    DataSet_outputFolder, DataSet_FileName,...
    Cluster_outputFolder, outputname);

save([DataSet_outputFolder, '\NumAccumulaNewDataSet.mat'], 'n_AccumulaNewDataSet')