Layer image with opacity over the top of another image. - OpenCV

Question

Edit

Anybody with a similar problem - I found another SO answer here with a great python solution that exploits the speed of NumPy.

Please consider the following problem:

I have two images, both the same size. One is a red square with varying layers of opacity:

And a second, a blue square, smaller than the red, with no opacity but white surrounding it.

I am using OpenCV's python bindings for this project and so far (after reading about watermarking here I have this:

redSquare = cv2.imread('redSquare.png', cv2.IMREAD_UNCHANGED)
(rH, rW) = redSquare.shape[:2]

blueSquare = cv2.imread('blueSquare.png')
(h, w) = blueSquare.shape[:2]

blueSquare = np.dstack([blueSquare, np.ones((h,w), dtype = 'uint8') * 255])
overlay = np.zeros((h,w,4), dtype = 'uint8')
overlay[0:rH, 0:rW] = redSquare
output = blueSquare .copy()
cv2.addWeighted(overlay, 0.5, output, 0.5, 0, output)

cv2.imwrite('imageAdded.png', output)

Which produces the following output:

However the desired effect is:

Now I understand by using weighted adding, I am using 0.5 of each, when I really want 1.0 of each, however when I try increasing the weighht of both, only one is increased and the other is decreased.

If anyone has some insight into how I can achieve this, preferably in Python, but if you know a way in C++ I am sure I can replicate it.

Thanks.

Abandon

please provide the red image with alpha channel. – Tussah

Sure, give me a sec and I will edit. – Abandon

Answer 1

here is C++ code gives exactly the result you want.

// http://jepsonsblog.blogspot.be/2012/10/overlay-transparent-image-in-opencv.html
// https://gist.github.com/maximus5684/082f8939edb6aed7ba0a

#include "opencv2/imgproc.hpp"
#include "opencv2/highgui.hpp"
#include "iostream"

using namespace cv;
using namespace std;

void overlayImage(Mat* src, Mat* overlay, const Point& location)
{
    for (int y = max(location.y, 0); y < src->rows; ++y)
    {
        int fY = y - location.y;

        if (fY >= overlay->rows)
            break;

        for (int x = max(location.x, 0); x < src->cols; ++x)
        {
            int fX = x - location.x;

            if (fX >= overlay->cols)
                break;

            double opacity = ((double)overlay->data[fY * overlay->step + fX * overlay->channels() + 3]) / 255;

            for (int c = 0; opacity > 0 && c < src->channels(); ++c)
            {
                unsigned char overlayPx = overlay->data[fY * overlay->step + fX * overlay->channels() + c];
                unsigned char srcPx = src->data[y * src->step + x * src->channels() + c];
                src->data[y * src->step + src->channels() * x + c] = srcPx * (1. - opacity) + overlayPx * opacity;
            }
        }
    }
}

int main( int argc, char** argv )
{
    Mat overlay = imread("ZuWDz.png",IMREAD_UNCHANGED);
    Mat underlay = imread("CtBAe.png",IMREAD_UNCHANGED);

    if( underlay.empty() || overlay.empty() )
    {
        return -1;
    }

    overlayImage( &underlay, &overlay, Point() );
    imshow("underlay result",underlay);

    waitKey();

    return 0;
}

Tussah

i can also show how to add this code to OpenCV library and call from Python if you able to rebuild OpenCV. – Tussah

However I am using OpenCV 3.1.0, as I notice that appears to be including opencv2, will that cause any issues? – Abandon

i will soon make a Pull Request to add this feature to OpenCV. i hope it will be accepted and merged. otherwise you can add your OpenCV source and rebuild library. – Tussah

Great, I hope it gets accepted as well. Pretty useful! – Abandon

here you can find how add the functionality to OpenCV. i think it is too primitive to be merged in the library but i shared it to show the way. – Tussah

@Tussah Did you pull request that in the end? – Cressida