Dionysos's profile - activity

Two of my threads need to read/write into the same Mat declared as global variable. Is Mat thread-safe or should I take care of it ? What can you recommend me ? Thanks

Steve, that's what I thought but I was using a binary image (the top one). The bottom one is the contours drawn (where you see that there's too many).

I found the problem. It wasn't using CV_RETR_TREE. findContours only works properly on "edge images". So you have to run canny first

Is it normal that find contours finds so many contours where there are obviously only 3 contours ?

The code is :

Mat image;
    vector<vector<Point> > contours;
    vector<Vec4i> hierarchy;
    image = imread(argv[1], 0);   // Read the file
    findContours(image, contours, hierarchy, CV_RETR_TREE, CV_CHAIN_APPROX_SIMPLE, Point(0, 0));
    cout << contours.size();


    RNG rng(12345);
    Mat drawing = Mat::zeros(image.size(), CV_8UC3);
    for (int i = 0; i< contours.size(); i++)
    {
        Scalar color = Scalar(rng.uniform(0, 255), rng.uniform(0, 255), rng.uniform(0, 255));
        drawContours(drawing, contours, i, color, 2, 8, hierarchy, 0, Point());
    }
    /// Show in a window
    namedWindow("Contours", CV_WINDOW_AUTOSIZE);
    imshow("Contours", drawing);
namedWindow("Display window", WINDOW_AUTOSIZE);// Create a window for display.
    imshow("Display window", image);                   // Show our image inside it.

Try adaptive thresholding

Convert your image to HSV using the cvtcolor function before feeding it to inRange

This is in order to normalize the cv::Mat before feeding it to an ANN

I want to print it on one line in a file, because this is the format of the training files of the ANN I am using. Instead of two loops, is there some prebuilt function for that ?

I want to resize some images (that range from 20 by 20 pixel to 110 by 110 pixel) to 20*20 pixel (for ANN training). There are many options for cv:resize. What is the most accurate for my case ? (performance is of no concern here).