OpenCV HOWTO: capture camera and do linear transformation on image

In this tutorial I'll try to show you how to do magic with OpenCV. I hope you've already installed it. We'll need some standard namespaces:

	using namespace cv;
	using namespace std;

view raw cv usings hosted with ❤ by GitHub

First, we have to create some Mat objects in which we'll store our images:

	Mat_<Vec3b> image;
	Mat grayImage;
	Mat negativeImage(480, 640, CV_8UC1);
	Mat_<Vec3b> colorNegativeImage(480, 640, CV_16U);
	Mat_<Vec3b> manipulatedImage(480, 640, CV_16U);

view raw mats hosted with ❤ by GitHub

And variables to store linear transformation values:

	int value;
	double plus = 0.0, mult = 1.0;

view raw values hosted with ❤ by GitHub

Then we intialize capturing camera image:

	VideoCapture cap(0); // 0 for notebook camera, 1 for external camera
	if(!cap.isOpened())
	{
	cout << "Camera error";
	return -1;
	}
	cap.open(0);

view raw camera capture hosted with ❤ by GitHub

And open windows to show results:

	namedWindow("window", CV_WINDOW_AUTOSIZE);
	namedWindow("grayWindow", CV_WINDOW_AUTOSIZE);
	namedWindow("negativeGrayWindow", CV_WINDOW_AUTOSIZE);
	namedWindow("negativeColorWindow", CV_WINDOW_AUTOSIZE);
	namedWindow("linearWindow", CV_WINDOW_AUTOSIZE);

view raw open windows hosted with ❤ by GitHub

Then we have to open infinite loop. In each iteration we have to capture the image from camera:

	// original image
	cap >> image;

view raw original image hosted with ❤ by GitHub

We can also make it gray:

	// bw
	cvtColor(image, grayImage, CV_BGR2GRAY);

view raw gray image hosted with ❤ by GitHub

If we want to get negative image, we have to subtract the image pixels values from 255 (because 255 is maximum value):

	// bw negative
	for (int i = 0; i < image.rows; i++) {
	for (int j = 0; j < image.cols; j++) {
	negativeImage.at<uchar>(i,j) = 255 - grayImage.at<uchar>(i,j);
	}
	}

view raw bw negative hosted with ❤ by GitHub

If we want to get color negative image, we have to subtract the image pixels values in all dimensions (3) from 255:

	// color negative
	for (int i = 0; i < image.rows; i++) {
	for (int j = 0; j < image.cols; j++) {
	for (int k = 0; k < 3; k++) {
	colorNegativeImage.at<Vec3b>(i,j)[k] = 255 - image.at<Vec3b>(i,j)[k];
	}
	}
	}

view raw color negative hosted with ❤ by GitHub

For linear transformation we have to calculate each dimension of each pixel. If we have to increase (decrease) contrast, we have to multiply source value by value grater than 1 (smaller than 1). Adding (subtracting) value to the source gives increased (decreased) brightness:

	// image manipulation
	for (int i = 0; i < image.rows; i++) {
	for (int j = 0; j < image.cols; j++) {
	for (int k = 0; k < 3; k++) {
	value = image.at<Vec3b>(i,j)[k] * mult + plus;
	if (value > 255) {
	value = 255;
	}
	if (value < 0) {
	value = 0;
	}
	manipulatedImage.at<Vec3b>(i,j)[k] = value;
	}
	}
	}

view raw linear transform hosted with ❤ by GitHub

Next we want to show our images:

	imshow("window", image);
	imshow("grayWindow", grayImage);
	imshow("negativeGrayWindow", negativeImage);
	imshow("negativeColorWindow", colorNegativeImage);
	imshow("linearWindow", manipulatedImage);

view raw show windows hosted with ❤ by GitHub

At the end, we want to manipulate the values to linear transformation by pressing arrows or pressing ESC to leave the application.

	int c = waitKey(20);
	switch(c)
	{
	case 27: //ESC
	destroyAllWindows();
	return 0;
	break;
	case 2621440: //down
	plus -= 20;
	break;
	case 2490368: //up
	plus += 20;
	break;
	case 2555904: //right
	mult += 0.1;
	break;
	case 2424832: //left
	mult -= 0.1;
	break;
	case 32: //space, reset parameters
	plus = 0;
	mult = 1;
	break;
	default:
	break;
	}

view raw linear params hosted with ❤ by GitHub

That's all, hope you'll find it useful!