OpenCV - How to convert grayscale to color (CV_8UC3)?

Question

I am trying to add two Mat objects together, but I am facing an error.

This is the working code:

    Mat src1, src2, dst;

    /// Read image ( same size, same type )
    src1 = imread("lion.png");
    src2 = imread("bluebell.png");

    dst = src1 + 0.5 * src2;

    imshow("Blend", dst);

    waitKey(0);
    return 0;

Both src1 and src2 have the same type which is CV_8UC3. But when I try this:

    Mat src1, src2, src3, dst;

    /// Read image
    src1 = imread("lion.png");
    src2 = imread("bluebell.png", IMREAD_GRAYSCALE);
    src2.convertTo(src3, COLOR_GRAY2RGB);

    cout << "src1.type " << src1.type() << endl;
    cout << "src2.type " << src2.type() << endl;
    cout << "src3.type " << src3.type() << endl;

    dst = src1 + 0.5 * src3;

    imshow("Blend", dst);

    waitKey(0);
    return 0;

It doesn't work. Because src2 and src3 both have type 0 which is CV_8U. But I want src3 to have type 16 which is CV_8UC3 (according to here). This code brings this error:

OpenCV Error: Sizes of input arguments do not match (The operation is neither 'array op array' (where arrays have the same size and the same number of channels), nor 'array op scalar', nor 'scalar op array') in cv::arithm_op

How can I convert a grayscale image to a colored one with exact type of CV_8UC3?

A grayscale image contains no colour information. How would you know what colour to map to what gray tone? — Jesper Juhl, Jan 07 '20 at 16:51
@JesperJuhl this is true. But I want `src3` to be `CV_8UC3` so that it can be added with `src1`. Even when I initialize `src3` to `CV_8UC3` doesn't work. — Hadi GhahremanNezhad, Jan 07 '20 at 16:54

rafix07 · Accepted Answer · 2020-01-07T17:35:09.647

You cannot change number of channels by calling convertTo. According to reference:

desired output matrix type or, rather, the depth since the number of channels are the same as the input has; if rtype is negative, the output matrix will have the same type as the input.

When you have matrix with one channel in grayscale, you can use cv::merge to create 3 channels image by putting this one component for BGR channels in new matrix:

src1 = imread("lion.png");
src2 = imread("bluebell.png", IMREAD_GRAYSCALE);    // one channel

cv::merge(std::vector<cv::Mat>{src2, src2, src2}, src2);
                            // blue, green,red    as output 3-channels mat
dst = src1 + 0.5 * src2;

By calling convertTo, you can convert data type of storing values, for example from integer to float.

Or you just use [`cvtColor`](https://docs.opencv.org/4.1.1/d8/d01/group__imgproc__color__conversions.html#ga397ae87e1288a81d2363b61574eb8cab): `cv::cvtColor(src2, src2, cv::COLOR_GRAY2BGR);`. — HansHirse, Jan 08 '20 at 05:34

OpenCV - How to convert grayscale to color (CV_8UC3)?

1 Answers1