main.cpp

/*
 * DetectLabel.h
 *
 *  Created on: May 1, 2014
 *      Author: chd
 */

#include <opencv2/opencv.hpp>
#include <opencv2/highgui.hpp>
#include <opencv2/ml.hpp>

#include <iostream>
#include <math.h>
#include <string.h>
#include <time.h>

#include "DetectLabel.h"
#include "LabelOCR.h"

using namespace cv;
using namespace std;


int main( int argc, char** argv )
{
    (void)argc;
    (void)argv;

    VideoCapture cap(0); // open the default camera
    if(!cap.isOpened())  // check if we succeeded
        return (-1);

    Mat normalImage, modImage, cropImage1, labelImage1;
    Mat cropImage2, labelImage2, binImage;
    vector<Point> contour;
    vector<vector<Point> > contours;
    Rect label1ROI;

    string text1, text2;

    DetectLabel detectLabels;
    LabelOCR labelOcr;
	Ptr< cv::ml::SVM > svmClassifier = cv::ml::SVM::create();

    vector<Mat> possible_labels, label_1, label_2;
    vector<string> labelText1, labelText2;
    detectLabels.showBasicImages = true;
    detectLabels.showAllImages = true;

    namedWindow("normal",WINDOW_NORMAL);

    // SVM learning algorithm
    clock_t begin_time = clock();
    // Read file storage.
    FileStorage fs;
    fs.open("/home/chd/Documents/sandbox/openCV_Tesseract_test/ml/SVM.xml", FileStorage::READ);
    Mat SVM_TrainingData;
    Mat SVM_Classes;
    fs["TrainingData"] >> SVM_TrainingData;
    fs["classes"] >> SVM_Classes;

    /* Set values to train SVM */
    svmClassifier->setCoef0( 0.0 );
    svmClassifier->setDegree( 0 );
    svmClassifier->setTermCriteria( TermCriteria(TermCriteria::MAX_ITER + TermCriteria::EPS, 1000, 0.01 ) );
    svmClassifier->setGamma( 0 );
    svmClassifier->setKernel( cv::ml::SVM::LINEAR );
    svmClassifier->setNu( 0 );
    svmClassifier->setP( 1.0 ); // for EPSILON_SVR, epsilon in loss function?
    svmClassifier->setC( 0.1 ); // From paper, soft classifier
    svmClassifier->setType( cv::ml::SVM::C_SVC ); // C_SVC; // EPSILON_SVR; // may be also NU_SVR; // do regression task
    svmClassifier->train( SVM_TrainingData, cv::ml::ROW_SAMPLE, SVM_Classes );

    float timer = ( clock () - begin_time ) /  CLOCKS_PER_SEC;
    cout << "Time: " << timer << endl;

    while(true)
    {
        cap >> normalImage; // get a new frame from camera
        imshow("normal", normalImage);

        possible_labels.clear();
        label_1.clear();
        label_2.clear();

        // segmentation
        detectLabels.segment(normalImage,possible_labels);

        int posLabels = possible_labels.size();
        if (posLabels > 0)
        {
            //For each possible label, classify with svm if it is a label or no
            for(int i=0; i< posLabels; i++)
            {
                if (!possible_labels[i].empty() )
                {
                    Mat gray;
                    cvtColor(possible_labels[i], gray, cv::COLOR_RGB2GRAY);
                    Mat p= gray.reshape(1, 1);
                    p.convertTo(p, CV_32FC1); 
                    int response = (int)svmClassifier->predict( p );
                    cout << "Class: " << response << endl;
                    if(response==1)
                        label_1.push_back(possible_labels[i]);
                    if(response==2)
                        label_2.push_back(possible_labels[i]);
                }
            }
        }
        if ( label_1.size() > 0) 
        {
            labelText1 = labelOcr.runRecognition(label_1,1);
        }
        if ( label_2.size() > 0) {
            labelText2 = labelOcr.runRecognition(label_2,2);
        }

        if(waitKey(30) >= 0) break;
    }

    return (0);
}