opencv video analysing

edit

You can threshold your image with a relative low threshold and than count the non zero pixels, which should be under a threshold because of bad pixels and camera noise. Example:

threshold(imgIn, imgOut, 5, 255, THRESH_BINARY);
if(countNonZero(imgOut) < maxNonZeroPixelsInBlackImage)
    cout << "Image is a black one!" << endl;

Or you make an inverse threshold, than findContours and if the area of your found contour is nearly as big as the image area it is a black one. Remember noise and bad pixels so it must be a bit smaller

vector<Mat> Contours;
threshold(imgIn, imgOut, 5, 255, THRESH_BINARY_INV);
findContours(imgOut, Contours, RETR_EXTERNAL, CHAIN_APPROX_SIMPLE, Point(0, 0));
for(int i = 0; i < Contours.size(); i++) {
    if(contourArea(Contours[i]) > imgIn.total()*0.99) {
        cout << "Image is a black one!" << endl;
        break;
    }
}

The first one should be the easier method, the second one should be more flexible, if just a part of the image has to be black or if you search a special area it can be adjusted. Check out: contours in OpenCV

You could use minMaxLoc on the frame and then check if the max value is 0. If so, you could issue the alert.