OpenCV: A straighforward method of colorizing a grayscale image

Question

What is a straight forward method of "colorizing" a grayscale image. By colorizing, I mean porting the grayscale intensity values to one of the three R, G, B channels in a new image.

For example a 8UC1 grayscale pixel with intensity of I = 50 should become a 8UC3 color pixel of intensity BGR = (50, 0, 0) when the picture is colorized to "blue".

In Matlab for example, what I'm asking for can be simply created with two lines of code:

color_im = zeros([size(gray_im) 3], class(gray_im));
color_im(:, :, 3) = gray_im; 

But amazingly I cannot find anything similar in OpenCV.

Answer 1

Well, the same thing requires a bit more work in C++ and OpenCV:

// Load a single-channel grayscale image
cv::Mat gray = cv::imread("filename.ext", CV_LOAD_IMAGE_GRAYSCALE);

// Create an empty matrix of the same size (for the two empty channels)
cv::Mat empty = cv::Mat::zeros(gray.size(), CV_8UC1);

// Create a vector containing the channels of the new colored image
std::vector<cv::Mat> channels;

channels.push_back(gray);   // 1st channel
channels.push_back(empty);  // 2nd channel
channels.push_back(empty);  // 3rd channel

// Construct a new 3-channel image of the same size and depth
cv::Mat color;
cv::merge(channels, color);

or as a function (compacted):

cv::Mat colorize(cv::Mat gray, unsigned int channel = 0)
{
    CV_Assert(gray.channels() == 1 && channel <= 2);

    cv::Mat empty = cv::Mat::zeros(gray.size(), gray.depth());
    std::vector<cv::Mat> channels(3, empty);
    channels.at(channel) = gray;

    cv::Mat color;
    cv::merge(channels, color);
    return color;
}