Features2d detect

[fen1ksss 0]

Hi guys, do anybody know, how the detect-function works? for example here:

Mat img = imread (...);    

    SurfFeatureDetector detector(400); 

    vector<KeyPoint> keypoints;

    detector.detect(img, keypoints);

especially how it's writting in keypoints1 .

[Smorodov 579]

Fragment from opencv\modules\features2d\src\keypoint.cpp

I think it looks quite self-explaining:

void write(FileStorage& fs, const string& objname, const vector<KeyPoint>& keypoints)

{

    WriteStructContext ws(fs, objname, CV_NODE_SEQ + CV_NODE_FLOW);


    int i, npoints = (int)keypoints.size();

    for( i = 0; i < npoints; i++ )

    {

        const KeyPoint& kpt = keypoints[i];

        write(fs, kpt.pt.x);

        write(fs, kpt.pt.y);

        write(fs, kpt.size);

        write(fs, kpt.angle);

        write(fs, kpt.response);

        write(fs, kpt.octave);

        write(fs, kpt.class_id);

    }

}

[/code]


Look also to other files in folder opencv\modules\features2d\src\.


Example of using  descriptors you can find in samples folder in file

descriptor_extractor_matcher.cpp


If you need SURF descriptor explanation, there a lot of documents in internet (http://en.wikipedia.org/wiki/SURF) for example.


May be I don't understand you correctly?