java - opencv blend (overlay) two image with different channels

Question

I want to overlay two images (maybe different format (channels) with opencv. Originally I use addWeighted, however, it fails on two images with different channels. Is there any way in opencv that can handle this case? Thanks

Answer 1

No, there is no API in OpenCV that offers this feature nativelly. On the other hand, there's nothing stopping you from writing your own code to do that.

A few weeks ago a made some changes to a function I saw somewhere else on the Internet to be able to:

• Pass two input images: the background as BGR and the foreground as BGRA;
• Blend them together according to a simple transparency rule;

(I don't remember where most of this code came from, sorry... but thank you whoever you are!)

void overlayImage(const cv::Mat &background, const cv::Mat &foreground, cv::Mat &output, cv::Point2i location, double opacity = 1.0)
{
    background.copyTo(output);

    // start at the row indicated by location, or at row 0 if location.y is negative.
    for (int y = std::max(location.y , 0); y < background.rows; ++y) {
        int fY = y - location.y; // because of the translation

        // we are done of we have processed all rows of the foreground image.
        if (fY >= foreground.rows)
            break;

        // start at the column indicated by location, or at column 0 if location.x is negative.
        for (int x = std::max(location.x, 0); x < background.cols; ++x) {
            int fX = x - location.x; // because of the translation.

            // we are done with this row if the column is outside of the foreground image.
            if (fX >= foreground.cols)
                break;

            // determine the opacity of the foregrond pixel, using its fourth (alpha) channel.
            double opacity_level = ((double)foreground.data[fY * foreground.step + fX * foreground.channels() + 3]) / 255.;
            if (opacity >= 0.0 && opacity < 1.0)
                opacity_level *= opacity;

            // and now combine the background and foreground pixel, using the opacity, but only if opacity > 0.
            for (int c = 0; opacity_level > 0 && c < output.channels(); ++c) {
                unsigned char foregroundPx = foreground.data[fY * foreground.step + fX * foreground.channels() + c];
                unsigned char backgroundPx = background.data[y * background.step + x * background.channels() + c];
                output.data[y*output.step + output.channels()*x + c] = backgroundPx * (1.-opacity_level) + foregroundPx * opacity_level;
            }
        }
    }
}

Below you will find the input images used for testing: the left is the background and the image on the right is the foreground.

To copy the foreground completely over the background, just do:

cv::Mat background = cv::imread("road.png");         // 3-chan BGR
cv::Mat foreground= cv::imread("tulip.png", -1);     // 4-chan BGRA
cv::Point location(0, 0);  

cv::Mat output;
overlayImage(input_bkg, input_target, output, location, 1.0);
cv::imwrite("output_alpha1.0.png", output);

and to perform a copy with 50% transparency:

overlayImage(input_bkg, input_target, output, location, 0.5);
cv::imwrite("output_alpha0.5.png", output);

Here are the results:

This implementation is not bulletproof for production purposes, so use it at your own risk.