Ask Your Question
0

Classification of Image

asked 2016-01-30 14:42:59 -0500

Rah1 gravatar image

1st of all I am very new to OpenCV as well as computer programming so sorry for asking a very basic and easy question but, I was not able to figure out how to do this.I want to classify blood samples images and want to determine blood group of a person. In order to do it, I have 1 reference image and test image and I want to check whether the sample image is similar to that of a reference image. if the image is same it should give me value "1" or"True" otherwise, it should give me " 0" or "False". how can I do that I am doing it in c++ and visual studio? I tried searching and found it can be done using LBP but the code was in Python and I don't know Python at all so was not able to figure it out.!

Reference image:

.image description

with

test1.image description

test2.image description

edit retag flag offensive close merge delete

Comments

1
  • what exactly are you looking for in the images, to determine a bloodgroup ?
  • if you want to train some classifier, you will needs lots of data. a single reference image won't do.
  • even if you don't understand the python code, it would be nice , to see it, maybe someone has an idea about it.
berak gravatar imageberak ( 2016-01-31 01:45:33 -0500 )edit

yup, I am looking to determine blood group.I am looking for whether the image is agglutinated or not you can see that reference image and test2 image are agglutinated while test1 image is not I want to see whether the image is agglutinated or not. if there is a better alternative than classifier that will also work I hope you got what I am looking for in an image. and thanks berak :)

Rah1 gravatar imageRah1 ( 2016-01-31 05:18:17 -0500 )edit

1 answer

Sort by ยป oldest newest most voted
4

answered 2016-02-01 03:18:39 -0500

berak gravatar image

well, maybe it even works with lbp histograms, the procedure would work like this:

  1. gather a lot of (grayscale) train images for each case.
  2. make an lbp histogram for each image, and add this histogram, and a label (agglutinated or not) to the training dataset.
  3. setup a classifier with this data. compareHist() is used with 1-nearest-neighbour classification below.
  4. then, later, you can predict agglutination on an image, 1st take the histogram again, then feed it to the classifier for prediction.

here's some code:

#include "opencv2/opencv.hpp"
#include <iostream>

using namespace cv;
using namespace std;

//
// if you squint hard, you'll see the close similarity to opencv's face recognition code ;)
//
void lbp_hist(const Mat &I, Mat &histogram)
{
    Mat_<uchar> img(I);
    Mat_<float> hist(1, 256, 0.0f);
    const int m=1;
    for (int r=m; r<img.rows-m; r++)
    {
        for (int c=m; c<img.cols-m; c++)
        {
            uchar v = 0;
            uchar cen = img(r,c);
            v |= (img(r-1,c  ) > cen) << 0;
            v |= (img(r-1,c+1) > cen) << 1;
            v |= (img(r  ,c+1) > cen) << 2;
            v |= (img(r+1,c+1) > cen) << 3;
            v |= (img(r+1,c  ) > cen) << 4;
            v |= (img(r+1,c-1) > cen) << 5;
            v |= (img(r  ,c-1) > cen) << 6;
            v |= (img(r-1,c-1) > cen) << 7;
            hist(v) ++;
        }
    }
    histogram = hist;
}

struct Classifier
{
    vector<Mat> histograms;
    vector<int> labels;

    Classifier( vector<Mat> &histograms, vector<int> &labels)
        : histograms(histograms)
        , labels(labels)
    {} // nothing else to do

    // nearest neighbour distance:
    int predict(const Mat &histogram)
    {
        double minDist=DBL_MAX;
        int minLabel = -1;
        for (size_t i=0; i<histograms.size(); i++)
        {
            // i tried a few, and CHISQR_ALT seemed best.
            double d = compareHist(histogram, histograms[i], HISTCMP_CHISQR_ALT);
            if (d < minDist)
            {
                minDist = d;
                minLabel = labels[i];
            }
        }
        return minLabel;
    }
};

int main() 
{
    Mat ref = imread("agref.jpg",1);
    Mat a1 = imread("ag1.jpg",1);
    Mat a2 = imread("ag2.jpg",1);
    // make histograms
    Mat h1,h2,h3;
    lbp_hist(ref,h1);
    lbp_hist(a1,h2);
    lbp_hist(a2,h3);
    // since we only got a few images(you need **lots* more!!)
    // augment our train data by simply flipping the existing images:
    Mat f1;  flip(ref,f1,0); // x-axis
    Mat f2;  flip(a1,f2,0);
    Mat f3;  flip(a2,f3,0);
    Mat h4,h5,h6;
    lbp_hist(f1,h4);
    lbp_hist(f2,h5);
    lbp_hist(f3,h6);

    // setup data for classifying:
    // (a histogram, and a label for each img)
    vector<Mat> hists;
    vector<int> labels; // 1==agglutinated, 0==not.
    // ref
    hists.push_back(h1); labels.push_back(0);
    hists.push_back(h4); labels.push_back(0);
    // agglutinated:
    hists.push_back(h2); labels.push_back(1);
    hists.push_back(h3); labels.push_back(1);
    hists.push_back(h5); labels.push_back(1);
    hists.push_back(h6); labels.push_back(1);

    // "train" the classifier:
    Classifier cls(hists, labels);

    // flip images for testing, so they again look a bit different:
    // (again, this is only for the demo)
    Mat f7;  flip(ref,f7,1); // y-axis
    Mat f8;  flip(a1,f8,1);
    Mat f9;  flip(a2,f9,1);
    Mat h7,h8,h9;
    lbp_hist(f7,h7);
    lbp_hist(f8,h8);
    lbp_hist(f9,h9);

    int p1 = cls.predict(h7); // ref
    int p2 = cls ...
(more)
edit flag offensive delete link more

Comments

Thanks, #Berak That works just the way I wanted :) but can u help me a bit more? how can I train images and what do you mean by train images I mean how can I do it? do I need to pass more images as reference images?

Rah1 gravatar imageRah1 ( 2016-02-04 12:49:28 -0500 )edit

did you understand, how it works ?

a nearest neighbour classifier actually does not need any training (opposed to using svm, ann or such) , all it needs is a copy of the data.

adding more images will improve the prediction result, get lots of images. 500 positives an 500 negatives will make a nice classification

again, in real life, you won't need the 'flip' part (that was more a way of getting more images), but rather have 'real' images instead.

berak gravatar imageberak ( 2016-02-04 13:12:01 -0500 )edit

okay thanks a lot berak :)

Rah1 gravatar imageRah1 ( 2016-02-04 13:39:02 -0500 )edit

berak a little more help plz.... what things should be added to the same code to run on GPU? thanks a lot

Rah1 gravatar imageRah1 ( 2016-05-13 15:49:56 -0500 )edit
Login/Signup to Answer

Question Tools

2 followers

Stats

Asked: 2016-01-30 14:05:23 -0500

Seen: 269 times

Last updated: Feb 01 '16