opencvImage Content Modification

Set Whole Image to a Solid Color

Given a non-empty cv::Mat img of some size, we can fill it to a solid color in several ways:

img = cv::Scalar(blueVal,greenVal,redVal);

or, the more general, mask supporting, cv::Mat::setTo():


If you are using the older OpenCV C API with IplImage* img:


cvSet(img, CV_RGB(redVal,greenVal,blueVal));

Pixel by pixel modification of images

In OpenCV, images can be RGB/BGR, HSV, grayscaled, black-white and so on. It is crucial to know the data type before dealing with images.

The image data types are mainly CV_8UC3 (Matrix of uchar with 3 channels) and CV_8U (Matrix of uchar with 1 channel), however, the conversion to other types such as CV_32FC3, CV_64F are also possible. (see data types)

Consider, the image is an RGB image which is read by imread function.

Mat rgb = imread('path/to/rgb/image', CV_LOAD_IMAGE_COLOR);
//to set RED pixel value of (i,j)th to X,<Vec3b>(i,j)[0] = X;

Similarly, if the image is grayscaled,<uchar>(i,j) = X;

Note that, in OpenCV, Black&White images are stored as CV_8U type with the values 0 and 255. Therefore, changing BW images are same as gray images.

Image color modification in OpenCV - kmeans(). To scan all the pixels of an image and replace the pixel values with generic colors.

#include opencv2/opencv.hpp> #include vector> using namespace std; using namespace cv; int main() { Mat3b img = imread("test.jpg");
imshow("Original", img);

// Cluster

int K = 8;
int n = img.rows * img.cols;
Mat data = img.reshape(1, n);
data.convertTo(data, CV_32F);

Mat labels;
Mat1f colors;
kmeans(data, K, labels, cv::TermCriteria(), 1, cv::KMEANS_PP_CENTERS, colors);

for (int i = 0; i < n; ++i)
{<float>(i, 0) = colors(<int>(i), 0);<float>(i, 1) = colors(<int>(i), 1);<float>(i, 2) = colors(<int>(i), 2);

Mat reduced = data.reshape(3, img.rows);
reduced.convertTo(reduced, CV_8U);

imshow("Reduced", reduced);

return 0;