OpenCV4 图像处理与视频分析实战教程笔记

14. 图像直方图的比较

void hist_compare() {
	Mat src1 = imread("D:/images/hist_01.jpg");
	Mat src2 = imread("D:/images/hist_02.jpg");

	Mat hist1, hist2;
	int channels[] = { 0, 1, 2 };
	int histSize[] = { 256, 256, 256 };
	float c1[] = { 0, 255 };
	float c2[] = { 0, 255 };
	float c3[] = { 0, 255 };
	const float* histRanges[] = { c1, c2, c3 };

	calcHist(&src1, 1, channels, Mat(), hist1, 3, histSize, histRanges, true, false);
	calcHist(&src2, 1, channels, Mat(), hist2, 3, histSize, histRanges, true, false);

	// 归一化
	normalize(hist1, hist1, 0, 1.0, NORM_MINMAX, -1, Mat());
	normalize(hist2, hist2, 0, 1.0, NORM_MINMAX, -1, Mat());

	double h12 = compareHist(hist1, hist2, HISTCMP_BHATTACHARYYA);
	double h11 = compareHist(hist1, hist1, HISTCMP_BHATTACHARYYA);
	cout << "HISTCMP_BHATTACHARYYA, h12= " << h12 << " h11=" << h11 << endl;

	double c12 = compareHist(hist1, hist2, HISTCMP_CORREL);
	double c11 = compareHist(hist1, hist1, HISTCMP_CORREL);
	cout << "HISTCMP_CORREL, c12= " << c12 << " c11=" << c11 << endl;

}

在这里插入图片描述

15. 使用LUT

void Advance::lut(Mat& src) {
	Mat color = imread("D:/images/lut.png");
	Mat lut = Mat::zeros(256, 1, CV_8UC3);
	for (size_t i = 0; i < 256; ++i) {
		lut.at<Vec3b>(i, 0) = color.at<Vec3b>(10, i);
	}
	imshow("color", color);
	Mat dst;
	LUT(src, lut, dst);
	imshow("lut", dst);
	applyColorMap(src, dst, COLORMAP_DEEPGREEN);
	imshow("default colormap", dst);
}

在这里插入图片描述

19. 自定义滤波

void Advance::filter2d(Mat& src) {
	// 均值滤波器
	int k = 15;
	Mat mkernel = Mat::ones(k, k, CV_32F) / (float)(k*k);
	Mat dst;
	filter2D(src, dst, -1, mkernel, Point(-1, -1), 0, BORDER_DEFAULT);
	imshow("2d filter", dst);

	// 非均值滤波
	Mat robot = (Mat_<int>(2, 2) << 1, 0, 0, -1);
	Mat result;
	filter2D(src, result, CV_32F, robot, Point(-1, -1), 127, BORDER_DEFAULT);
	convertScaleAbs(result, result);
	imshow("robot filter", result);
}

在这里插入图片描述

20. 图像的梯度

图像梯度的3个算子,robot,sobel和scharr算子
在这里插入图片描述

void Advance::gradient(Mat& src) {
	// robot 算子
	Mat robot_x = (Mat_<int>(2,2) << 1, 0, 0, -1);
	Mat robot_y = (Mat_<int>(2,2) << 0, 1, -1, 0);
	Mat grad_x, grad_y;
	filter2D(src, grad_x, CV_32F, robot_x, Point(-1, -1), 0, BORDER_DEFAULT);
	filter2D(src, grad_y, CV_32F, robot_y, Point(-1, -1), 0, BORDER_DEFAULT);
	convertScaleAbs(grad_x, grad_x);
	convertScaleAbs(grad_y, grad_y);
	Mat result;
	add(grad_x, grad_y, result);
	imshow("robot gradient", result);

	// Sobel算子
	Sobel(src, grad_x, CV_32F, 1, 0);
	Sobel(src, grad_y, CV_32F, 0, 1);
	convertScaleAbs(grad_x, grad_x);
	convertScaleAbs(grad_y, grad_y);
	Mat result2;
	addWeighted(grad_x, 0.5, grad_y, 0.5, 0, result2);
	imshow("sobel, add weight", result2);
	
	// Scharr算子
	Mat result3;
	Scharr(src, grad_x, CV_32F, 1, 0);
	Scharr(src, grad_y, CV_32F, 0, 1);
	convertScaleAbs(grad_x, grad_x);
	convertScaleAbs(grad_y, grad_y);
	addWeighted(grad_x, 0.5, grad_y, 0.5, 0, result3);
	imshow("scharr", result3);
}

几种不同算子的梯度提取效果.
在这里插入图片描述

21. 拉普拉斯算子和图像的锐化

拉普拉斯算子对图像的噪声比较敏感

void Advance::laplacian(Mat& src) {
	Mat dst;
	// 拉普拉斯二阶导梯度提取,对噪声很敏感
	Laplacian(src, dst, -1, 3, 1.0, 0, BORDER_DEFAULT);
	imshow("Laplacian demo", dst);

	// 图像锐化
	Mat dst2;
	Mat filter = (Mat_<int>(3, 3) << 0, -1, 0, 
									 -1, 5, -1, 
									  0, -1, 0);
	filter2D(src, dst2, CV_32F, filter, Point(-1, -1), 0, BORDER_DEFAULT);
	convertScaleAbs(dst2, dst2);
	imshow("sharpen demo", dst2);

}

在这里插入图片描述

22. USM锐化

sharp_image = blur - laplacian

void Advance::USM(Mat& src) {
	Mat blur;
	GaussianBlur(src, blur, Size(3, 3), 0);

	Mat lap;
	Laplacian(src, lap, -1, 1, 1.0, 0, BORDER_DEFAULT);

	Mat usm;
	addWeighted(blur, 1.0, lap, -1.0, 0, usm);

	namedWindow("usm", WINDOW_AUTOSIZE);
	imshow("usm", blur);
}

23. 图像噪声

噪声分为椒盐噪声和高斯噪声

void Advance::Noise(Mat& src) {
	Mat src2 = src.clone();

	RNG rng(12345);
	// 椒盐噪声
	int nums = 10000;
	int w = src.cols;
	int h = src.rows;

	cout << w << " " << h << endl;

	for (int i = 0; i < nums; ++i) {

		int x = rng.uniform(0, w);
		int y = rng.uniform(0, h);

		if (i % 2 == 0) {
			src.at<Vec3b>(y, x) = Vec3b(0, 0, 0);
		}
		else {
			src.at<Vec3b>(y, x) = Vec3b(255, 255, 255);
		}
	}
	imshow("椒盐噪声", src);


	Mat dst;
	Mat noise = Mat::zeros(src2.size(), src2.type());
	randn(noise, Scalar(25, 25, 25), Scalar(10, 10, 10));
	imshow("高斯噪声", noise);
	addWeighted(src2, 1.0, noise, 1.0, 0, dst);
	imshow("高斯噪声加入结果", dst);

}

在这里插入图片描述

24. 图像去噪

去噪有中值滤波,高斯滤波

void Advance::median_blur(Mat& src) {
	RNG rng(12345);
	// 椒盐噪声
	int nums = 10000;
	int w = src.cols;
	int h = src.rows;

	cout << w << " " << h << endl;

	for (int i = 0; i < nums; ++i) {

		int x = rng.uniform(0, w);
		int y = rng.uniform(0, h);

		if (i % 2 == 0) {
			src.at<Vec3b>(y, x) = Vec3b(0, 0, 0);
		}
		else {
			src.at<Vec3b>(y, x) = Vec3b(255, 255, 255);
		}
	}
	imshow("椒盐噪声", src);
	Mat dst;
	medianBlur(src, dst, 5);
	imshow("中值滤波去噪声", dst);

	GaussianBlur(src, dst, Size(5,5), 0);
	imshow("高斯滤波去噪", dst);

}

高斯滤波对椒盐噪声去除效果并没有中值滤波好
在这里插入图片描述

void Advance::median_blur(Mat& src) {
	
	Mat noise = Mat::zeros(src.size(), src.type());
	randn(noise, Scalar(50, 60, 70), Scalar(55, 50, 50));
	addWeighted(src, 1.0, noise, 0.7, 0, src);
	imshow("添加高斯噪声", src);

	Mat dst;
	medianBlur(src, dst, 5);
	imshow("中值滤波去噪声", dst);

	GaussianBlur(src, dst, Size(5, 5), 0);
	imshow("高斯滤波去噪", dst);

}

高斯噪声去除比椒盐噪声要困难很多,两种滤波的去除效果都很一般
在这里插入图片描述

25. 边缘保留滤波

分为高斯双边滤波和非局部均值滤波,其中非局部均值滤波运行很慢,灰度和彩色的api是分开的.

void Advance::bilateral_filter(Mat& src) {

	Mat noise = Mat::zeros(src.size(), src.type());
	randn(noise, Scalar(50, 60, 70), Scalar(55, 50, 50));
	addWeighted(src, 1.0, noise, 0.7, 0, src);
	imshow("添加高斯噪声", src);

	// 高斯双边滤波
	Mat dst;
	bilateralFilter(src, dst, 0, 200, 0);
	imshow("高斯双边滤波", dst);

	// 非局部均值滤波
	// 灰度图有效
	cvtColor(src, src, COLOR_BGR2GRAY);
	fastNlMeansDenoising(src, dst, 15, 10, 30);
	imshow("非局部均值滤波", dst);

}

在这里插入图片描述

26. canny边缘提取加上trackbar演示

static void canny_demo(int thres, void* user_data) {
	Mat src = *((Mat *)user_data);
	Mat edges;
	Canny(src, edges, thres, thres * 3, 3, false);
	imshow("canny output", edges);
}


void Advance::canny(Mat& src) {
	namedWindow("canny", WINDOW_AUTOSIZE);
	int thres=50;
	createTrackbar("canny threshod: ", "canny", &thres, 100, canny_demo, (void*)(&src));
	imshow("canny", src);
	canny_demo(thres, &src);
}

在这里插入图片描述

27. 二值图像的概念

五种二值分割方法
在这里插入图片描述

void Advance::threshold_demo(Mat& src) {
	Mat gray, binary;
	cvtColor(src, gray, COLOR_BGR2GRAY);
	threshold(gray, binary, 127, 255, THRESH_BINARY);
	imshow("THRESH_BINARY", binary);

	threshold(gray, binary, 127, 255, THRESH_BINARY_INV);
	imshow("THRESH_BINARY_INV", binary);

	threshold(gray, binary, 127, 255, THRESH_TRUNC);
	imshow("THRESH_TRUNC", binary);

	threshold(gray, binary, 127, 255, THRESH_TOZERO);
	imshow("THRESH_TOZERO", binary);

	threshold(gray, binary, 127, 255, THRESH_TOZERO_INV);
	imshow("THRESH_TOZERO_INV", binary);

}

在这里插入图片描述

28. 全局阈值

在这里插入图片描述

void Advance::global_threshold(Mat& src) {
	Mat gray, binary;
	cvtColor(src, gray, COLOR_BGR2GRAY);
	// 求均值
	Scalar m = mean(gray);
	threshold(gray, binary, m[0], 255, THRESH_BINARY);
	imshow("mean", binary);

	// OTSU
	double t1 = threshold(gray, binary, 0, 255, THRESH_BINARY | THRESH_OTSU);
	imshow("OTSU", binary);
	cout << "OTSU threshold: " << t1 << endl;
	   
	// 三角法 triangle
	double t2 = threshold(gray, binary, 0, 255, THRESH_BINARY | THRESH_TRIANGLE);
	imshow("triangle", binary);
	cout << "triangle threshold: " << t2 << endl;

}

在这里插入图片描述

29. 自适应阈值

自适应阈值主要用在有阴影的图片上面
在这里插入图片描述

void Advance::adjust_threshold(Mat& src) {
	Mat dst, gray;
	cvtColor(src, gray, COLOR_BGR2GRAY);

	adaptiveThreshold(gray, dst, 255, ADAPTIVE_THRESH_GAUSSIAN_C, THRESH_BINARY, 25, 10);
	imshow("ADAPTIVE_THRESH_GAUSSIAN_C", dst);

	adaptiveThreshold(gray, dst, 255, ADAPTIVE_THRESH_MEAN_C, THRESH_BINARY, 25, 10);
	imshow("ADAPTIVE_THRESH_MEAN_C", dst);
}

30. 联通件的扫描

void Advance::connect(Mat& src) {
	Mat gray, binary;
	GaussianBlur(src, src, Size(3, 3), 0);
	cvtColor(src, gray, COLOR_BGR2GRAY);
	threshold(gray, binary, 0, 255, THRESH_BINARY | THRESH_OTSU);
	imshow("otsu", binary);

	Mat labels = Mat::zeros(binary.size(), CV_32S);
	int num_labels = connectedComponents(binary, labels, 8, CV_32S, CCL_DEFAULT);
	printf("num of objects %d", num_labels-1);


	RNG rng(123);
	vector<Vec3b> colors(num_labels);
	colors[0] = Vec3b(0, 0, 0);
	for (int i = 1; i < num_labels; ++i) {
		colors[i] = Vec3b(rng.uniform(0, 255), rng.uniform(0, 255), rng.uniform(0, 255));
	}

	Mat result = Mat::zeros(src.size(), src.type());
	for (int h = 0; h < binary.rows; ++h) {
		for (int w = 0; w < binary.cols; ++w) {
			int label = labels.at<int>(h, w);
			result.at<Vec3b>(h, w) = colors[label];
		}
	}
	putText(result, format("num of objects %d", num_labels - 1), Point(50, 50), FONT_HERSHEY_PLAIN, 2, Scalar(0, 255, 0), 2, 8);
	imshow("colorful", result);

}

在这里插入图片描述

31. 带统计信息联通件扫描

统计的信息有联通件对中心，外接矩形的左上角坐标，高宽值，联通件面积等值。

void Advance::ccl_statistic_demo(Mat &image) {
	Mat gray, binary;
	cvtColor(image, gray, COLOR_BGR2GRAY);
	threshold(gray, binary, 0, 255, THRESH_BINARY | THRESH_OTSU);
	imshow("otsu", binary);

	Mat labels = Mat::zeros(binary.size(), CV_32S);
	Mat stats, centroids;
	int num_labels = connectedComponentsWithStats(binary, labels, stats, centroids, 8, CV_32S, CCL_DEFAULT);

	// 生成不同的颜色
	RNG rng(123);
	vector<Vec3b> colors(num_labels);
	colors[0] = Vec3b(0, 0, 0);
	for (int i = 1; i < num_labels; ++i) {
		colors[i] = Vec3b(rng.uniform(0, 255), rng.uniform(0, 255), rng.uniform(0, 255));
	}


	Mat result = Mat::zeros(image.size(), CV_8UC3);
	for (int h = 0; h < binary.rows; ++h) {
		for (int w = 0; w < binary.cols; ++w) {
			int label = labels.at<int>(h, w);
			result.at<Vec3b>(h, w) = colors[label];
		}
	}

	for (int i = 1; i < num_labels; ++i) {
		int cx = centroids.at<double>(i, 0);
		int cy = centroids.at<double>(i, 1);
		circle(result, Point(cx, cy), 3, Scalar(0, 255, 0), 2, LINE_8, 0);
		
		int x = stats.at<int>(i, CC_STAT_LEFT);
		int y = stats.at<int>(i, CC_STAT_TOP);
		int height = stats.at<int>(i, CC_STAT_HEIGHT);
		int width = stats.at<int>(i, CC_STAT_WIDTH);
		int area = stats.at<int>(i, CC_STAT_AREA);
		
		Rect rect(x, y, width, height);
		rectangle(result, rect, Scalar(0, 0, 255), 2, LINE_8, 0);
		putText(result, format("%d", area), Point(x, y), FONT_HERSHEY_PLAIN, 2, Scalar(255, 0, 0), 1, LINE_8);

	}

	putText(result, format("num of objects %d", num_labels - 1), Point(10, 10), FONT_HERSHEY_PLAIN, 1, Scalar(0, 255, 0), 2, 8);
	imshow("colorful", result);


}

在这里插入图片描述

32. 轮廓的发现和绘制

void Advance::find_contours(Mat &image) {

	Mat gray, binary;
	cvtColor(image, gray, COLOR_BGR2GRAY);
	threshold(gray, binary, 0, 255, THRESH_BINARY | THRESH_OTSU);
	imshow("otsu", binary);


	vector<vector<Point>> contours;
	vector<Vec4i> hirerachy;
	findContours(binary, contours, hirerachy, RETR_TREE, CHAIN_APPROX_SIMPLE, Point());
	drawContours(image, contours, -1, Scalar(0, 255, 0), 2, LINE_8);
	imshow("find contours", image);

}

在这里插入图片描述

33. 图像轮廓的计算

void Advance::cal_contours(Mat &image) {
	Mat gray, binary;
	cvtColor(image, gray, COLOR_BGR2GRAY);
	threshold(gray, binary, 0, 255, THRESH_BINARY_INV | THRESH_OTSU);
	imshow("otsu", binary);

	vector<vector<Point>> contours;
	vector<Vec4i> hirerachy;
	findContours(binary, contours, hirerachy, RETR_EXTERNAL, CHAIN_APPROX_SIMPLE, Point());
	
	for (size_t i = 0; i < contours.size(); ++i) {
		double area = contourArea(contours[i]);
		double len = arcLength(contours[i], true);

		if (area < 100 || area < 10) {
			continue;
		}
		Rect box = boundingRect(contours[i]);
		//drawContours(image, contours, i, Scalar(0, 255, 0), 2, LINE_8);
		//rectangle(image, box, Scalar(0, 255, 0), 2);
		RotatedRect rrt = minAreaRect(contours[i]);
		// 绘制最小包围的椭圆
		//ellipse(image, rrt, Scalar(255, 0, 0), 2);
		// 最小包围的斜矩形
		Point2f pts[4];
		rrt.points(pts);
		for (int j = 0; j < 4; ++j) {
			line(image, pts[j], pts[(j + 1) % 4], Scalar(255,0, 255), 2);
		}
		double angle = rrt.angle;
		printf("index: %d, area: %.1f, length: %.1f, angle: %.1f\n", i, area, len, angle);
	}
	imshow("find contours", image);
}

请添加图片描述

34. 轮廓匹配

void contour_info(Mat &image, vector<vector<Point>> &contours) {
	Mat dst, gray, binary;
	
	GaussianBlur(image, dst, Size(3, 3), 0);
	cvtColor(dst, gray, COLOR_BGR2GRAY);
	threshold(gray, binary, 0, 255, THRESH_BINARY | THRESH_OTSU);

	imshow("binary", binary);
	vector<Vec4i> hirearchy;
	findContours(binary, contours, hirearchy, RETR_EXTERNAL, CHAIN_APPROX_SIMPLE);

}


void Advance::match_contours() {
	Mat src1 = imread("D:/images/abc.png");
	Mat src2 = imread("D:/images/a5.png");
	if (src1.empty() || src2.empty()) {
		printf("could not load image");
		return;
	}
	imshow("input1", src1);
	imshow("imput2", src2);

	vector<vector<Point>> contours1, contours2;
	contour_info(src1, contours1);
	contour_info(src2, contours2);

	Moments mm2 = moments(contours2[0]);
	Mat hu2;
	HuMoments(mm2, hu2);
	for (size_t i = 0; i < contours1.size(); ++i) {
		Moments mm = moments(contours1[i]);
		double cx = mm.m10 / mm.m00;
		double cy = mm.m01 / mm.m00;
		circle(src1, Point(cx, cy), 3, Scalar(255, 0, 255), 2);
		Mat hu;
		HuMoments(mm, hu);
		double dist = matchShapes(hu, hu2, CONTOURS_MATCH_I1, 0);
		if (dist < 1.0) {
			printf("match distance value: %.2f\n", dist);
			drawContours(src1, contours1, i, Scalar(0, 255, 0), 2);
		}
	}
	imshow("contours match demo", src1);
}

请添加图片描述

3.5 轮廓逼近和拟合

使用线段去逼近

void Advance::approx_contours(Mat &image) {
	Mat dst;
	GaussianBlur(image, dst, Size(3, 3), 0);
	Mat gray;
	cvtColor(dst, gray, COLOR_BGR2GRAY);

	Mat binary;
	threshold(gray, binary, 0, 255, THRESH_BINARY | THRESH_OTSU);
	imshow("binary", binary);


	vector<vector<Point>> contours;
	vector<Vec4i> hirearchy;
	findContours(binary, contours, hirearchy, RETR_EXTERNAL, CHAIN_APPROX_SIMPLE);

	for (size_t i = 0; i < contours.size(); ++i) {
		Mat result;
		approxPolyDP(contours[i], result, 4, true);
		printf("coners: %d\n", result.rows);


		double c = arcLength(contours[i], true);
		double s = contourArea(contours[i]);

		// 计算中心点
		Moments mm = moments(contours[i]);
		double cx = mm.m10 / mm.m00;
		double cy = mm.m01 / mm.m00;
		if (result.rows == 3) {
			putText(image, format("triangle, c=%.0f, s=%.0f\n", c, s), Point(cx-10, cy-10), FONT_HERSHEY_PLAIN, 1, Scalar(255, 0, 255));
			circle(image, Point(cx, cy), 3, Scalar(0, 255, 0));
		}
	}
	imshow("approx contours", image);
}

请添加图片描述
使用椭圆去逼近

void Advance::approx_contours2(Mat &image) {
	Mat dst;
	GaussianBlur(image, dst, Size(3, 3), 0);
	Mat gray;
	cvtColor(dst, gray, COLOR_BGR2GRAY);

	Mat binary;
	threshold(gray, binary, 0, 255, THRESH_BINARY | THRESH_OTSU);
	imshow("binary", binary);

	vector<vector<Point>> contours;
	vector<Vec4i> hirearchy;
	findContours(binary, contours, hirearchy, RETR_EXTERNAL, CHAIN_APPROX_SIMPLE);
	//drawContours(image, contours, 0, Scalar(0, 255, 0), 2);

	RotatedRect rrt = fitEllipse(contours[0]);
	ellipse(image, rrt, Scalar(255, 0, 0), 2);

	imshow("approx contours", image);
}

请添加图片描述

36 霍夫直线检测

void Advance::hough_lines(Mat &image) {
	Mat dst;
	GaussianBlur(image, dst, Size(3, 3), 0);
	Mat gray;
	cvtColor(dst, gray, COLOR_BGR2GRAY);

	Mat binary;
	threshold(gray, binary, 0, 255, THRESH_BINARY | THRESH_OTSU);
	imshow("binary", binary);
	 
	// 霍夫直线检测
	vector<Vec3f> lines;
	HoughLines(binary, lines, 1, CV_PI / 180.0, 250, 0, 0);

	Point pt1, pt2;
	for (size_t i = 0; i < lines.size(); ++i) {
		float rho = lines[i][0]; // 距离
		float theta = lines[i][1]; //角度
		float acc = lines[i][2]; // 累加值
		

		double a = cos(theta);
		double b = sin(theta);
		double x0 = a * rho, y0 = b * rho;
		pt1.x = cvRound(x0 + 1000 * (-b));
		pt1.y = cvRound(y0 + 1000 * a);
		pt2.x = cvRound(x0 - 1000 * (-b));
		pt2.y = cvRound(y0 - 1000 * a);
		int angle = round(theta / CV_PI * 180);

		if (angle > 90) {
			line(image, pt1, pt2, Scalar(0, 0, 255), 2, 8, 0);
		}
		else {
			line(image, pt1, pt2, Scalar(0, 255, 0), 2, 8, 0);
		}
		printf("rho: %.2f, theta: %.2f, acc: %.2f, angle: %d\n", rho, theta, acc, angle);

	}
	imshow("hough lines", image);

}

在这里插入图片描述

37 霍夫直线检测2

void Advance::hough_linesp(Mat &image) {
	Mat binary;
	Canny(image, binary, 80, 160, 3, false);
	imshow("canny", binary);

	vector<Vec4i> lines;
	HoughLinesP(binary, lines, 1, CV_PI / 180, 80, 200, 10);
	Mat result = Mat::zeros(image.size(), image.type());
	for (int i = 0; i < lines.size(); ++i) {
		line(result, Point(lines[i][0], lines[i][1]), Point(lines[i][2], lines[i][3]), Scalar(0, 255, 0), 1);
	}
	imshow("hough linesp demo", result);
}

在这里插入图片描述

38. 霍夫圆检测

霍夫类型算法对噪声很敏感的，所以使用前需要先进行高斯模糊

void Advance::hough_circle(Mat &image) {
	Mat gray, binary;
	cvtColor(image, gray, COLOR_BGR2GRAY);
	GaussianBlur(gray, gray, Size(9, 9), 2, 2);
	imshow("gray", gray);

	vector<Vec3f> circles;
	int min_dis = 20, min_r = 10, max_r = 50;
	HoughCircles(gray, circles, HOUGH_GRADIENT, 2, min_dis, 100, 100, min_r, max_r);
	for (int t = 0; t < circles.size(); ++t) {
		Point centre(circles[t][0], circles[t][1]);
		double radius = round(circles[t][2]);
		circle(image, centre, radius, Scalar(0, 255, 0), 2);
	}
	imshow("hough circle", image);
}

有高斯模糊
在这里插入图片描述
没有高斯模糊

39. 图像形态学操作

膨胀和腐蚀

void Advance::erode_dilate(Mat &image) {
	Mat gray, binary;
	cvtColor(image, gray, COLOR_BGR2GRAY);
	threshold(gray, binary, 0, 255, THRESH_BINARY | THRESH_OTSU);
	Mat dst1, dst2;
	// 腐蚀
	Mat kernel = getStructuringElement(MORPH_RECT, Size(3, 3), Point(-1, -1));
	erode(binary, dst1, kernel);
	imshow("erode", dst1);
	// 膨胀
	dilate(binary, dst2, kernel);
	imshow("dilate", dst2);
}

在这里插入图片描述

40. 图像的开闭

void Advance::open_close(Mat &image) {
	// 开闭
	Mat gray, binary;
	cvtColor(image, gray, COLOR_BGR2GRAY);
	threshold(gray, binary, 0, 255, THRESH_BINARY | THRESH_OTSU);
	Mat dst1, dst2;
	// 闭操作 先膨胀后腐蚀，去掉黑色的小东西
	//Mat kernel = getStructuringElement(MORPH_ELLIPSE, Size(25, 25), Point(-1, -1));
	//morphologyEx(binary, dst1, MORPH_CLOSE, kernel, Point(-1, -1), 1);
	//imshow("open", dst1);

	// 开操作，先腐蚀后膨胀, 去掉白色（前景， OpenCV所有形态学操作，白是前景，黑色背景）的小东西
	Mat kernel1 = getStructuringElement(MORPH_RECT, Size(30, 1), Point(-1, -1));
	morphologyEx(binary, dst2, MORPH_OPEN, kernel1, Point(-1, -1));
	imshow("open", dst2);

}

闭操作：去掉黑点
请添加图片描述
开操作：去掉白点，保留横线

41. 形态学梯度

基本梯度 – 膨胀减去腐蚀之后的结果
内梯度 – 原图减去腐蚀后结果
外梯度 – 膨胀减去原图后结果

void Advance::morph_gradient(Mat &image) {
	Mat gray;
	cvtColor(image, gray, COLOR_BGR2GRAY);

	Mat kernel = getStructuringElement(MORPH_RECT, Size(3, 3), Point(-1, -1));

	
	Mat basic_grad;
	// 基本梯度 -- 膨胀减去腐蚀之后的结果
	morphologyEx(gray, basic_grad, MORPH_GRADIENT, kernel, Point(-1, -1));
	imshow("basic gradient", basic_grad);

	// 内梯度 -- 原图减去腐蚀后结果
	Mat dst1, inner_grad;
	morphologyEx(gray, dst1, MORPH_ERODE, kernel);
	subtract(gray, dst1, inner_grad);
	imshow("inner grad", inner_grad);

	// 外梯度 -- 膨胀减去原图后结果
	Mat dst2, outter_grad;
	morphologyEx(gray, dst2, MORPH_DILATE, kernel);
	subtract(dst2, gray, outter_grad);
	imshow("outter grad", outter_grad);

}

请添加图片描述

42. 更多形态学操作

// 顶帽：原图减去开操作之后的结果

void Advance::other_morph(Mat &image) {
	Mat gray, binary;
	cvtColor(image, gray, COLOR_BGR2GRAY);
	threshold(gray, binary, 0, 255, THRESH_BINARY_INV | THRESH_OTSU);
	imshow("binary", binary);

	// 顶帽： 原图减去开操作之后的结果
	Mat kernel = getStructuringElement(MORPH_ELLIPSE, Size(25, 25));
	Mat tophat, open;
	morphologyEx(binary, tophat, MORPH_TOPHAT, kernel);
	morphologyEx(binary, open, MORPH_OPEN, kernel, Point(-1, -1));
	imshow("open", open);
	imshow("tophat", tophat);

}

在这里插入图片描述
// 黑帽：闭操作之后结果减去原图

void Advance::other_morph(Mat &image) {
	Mat gray, binary;
	cvtColor(image, gray, COLOR_BGR2GRAY);
	threshold(gray, binary, 0, 255, THRESH_BINARY | THRESH_OTSU);
	imshow("binary", binary);

	// 黑帽：闭操作之后结果减去原图
	Mat kernel = getStructuringElement(MORPH_ELLIPSE, Size(25, 25));
	Mat blackhat, close;
	morphologyEx(binary, blackhat, MORPH_BLACKHAT, kernel);
	morphologyEx(binary, close, MORPH_CLOSE, kernel, Point(-1, -1));
	imshow("close", close);
	imshow("blackhat", blackhat);

}

在这里插入图片描述
hit and miss

void Advance::other_morph(Mat &image) {
	Mat gray, binary;
	cvtColor(image, gray, COLOR_BGR2GRAY);
	threshold(gray, binary, 0, 255, THRESH_BINARY | THRESH_OTSU);
	imshow("binary", binary);

	// hit and miss
	Mat kernel = getStructuringElement(MORPH_CROSS, Size(25, 25));
	Mat hitmiss;
	morphologyEx(binary, hitmiss, MORPH_HITMISS, kernel);
	imshow("hitmiss", hitmiss);
	

}

请添加图片描述

43. 综合练习找出卫星图岛屿轮廓

void Advance::image_process_demo(Mat &image) {
	Mat gray;
	cvtColor(image, gray, COLOR_BGR2GRAY);
	Mat blur;
	GaussianBlur(gray, blur, Size(3, 3), 0);
	Mat binary;
	// 二值化
	threshold(gray, binary, 0, 255, THRESH_BINARY | THRESH_OTSU);
	imshow("binary", binary);
	// 闭操作，用于填充最大的白色轮廓
	Mat morph;
	Mat kernel = getStructuringElement(MORPH_RECT, Size(15, 15), Point(-1, -1));
	morphologyEx(binary, morph, MORPH_CLOSE, kernel);
	imshow("morph", morph);

	vector<vector<Point>> contours;
	vector<Vec4i> hirerachy;
	findContours(morph, contours, RETR_EXTERNAL, CHAIN_APPROX_SIMPLE);

	Mat output = image.clone();
	double max_area = 0.;
	size_t max_index;
	for (size_t i = 0; i < contours.size(); ++i) {
		double area = contourArea(contours[i]);
		printf("index: %d, area: %.1f\n", i, area);
		if (area > max_area) {
			max_index = i;
			max_area = area;
		}
		drawContours(output, contours, i, Scalar(0, 255, 0), 2);
	}
	Mat result = Mat::zeros(image.size(), image.type());
	drawContours(result, contours, max_index, Scalar(0, 0, 255), 2);
	
	Mat pts;
	// 多边形逼近
	approxPolyDP(contours[max_index], pts, 4, true);
	for (int i = 0; i < pts.rows; ++i) {
		Vec2i pt = pts.at<Vec2i>(i, 0);
		circle(result, Point(pt[0], pt[1]), 2, Scalar(0, 255, 0), 2, 8, 0);
	}

	printf("max index: %d", max_index);
	imshow("output", output);
	imshow("result", result);
}

请添加图片描述

44. 视频读写api

读取一个视频信息并保存

void Advance::video_capture() {
	VideoCapture capture("D:/images/bike.avi");
	if (!capture.isOpened()) {
		printf("video is not opened");
		return;
	}
	namedWindow("frame", WINDOW_AUTOSIZE);

	int height = capture.get(CAP_PROP_FRAME_HEIGHT);
	int width = capture.get(CAP_PROP_FRAME_WIDTH);
	int fps = capture.get(CAP_PROP_FPS);
	int type = capture.get(CAP_PROP_FOURCC);
	printf("height: %d, width: %d, fps: %d", height, width, fps);

	VideoWriter writer("D:/test.mp4", type, fps, Size(width, height), true);
	   	 
	while (true) {
		char c = waitKey(50);
		if (c == 27) {
			break;
		}
		Mat frame;
		bool ret = capture.read(frame);
		if (!ret) {
			break;
		}
		writer.write(frame);
		imshow("frame", frame);
	}

	capture.release();
	writer.release();
	destroyAllWindows();
}

45. 色彩空间转换

使用色彩空间转换进行绿幕抠图

void Advance::color_space() {
	VideoCapture capture("D:/images/01.mp4");

	while (true) {
		char c = waitKey(50);
		if (c == 27) {
			break;
		}
		Mat frame, hsv, lab;
		bool ret = capture.read(frame);

		if (!ret) {
			break;
		}
		cvtColor(frame, hsv, COLOR_BGR2HSV);
		//cvtColor(frame, lab, COLOR_BGR2Lab);
		//imshow("hsv", hsv);
		//imshow("lab", lab);
		
		Mat mask,dst;
		inRange(hsv, Scalar(35, 43, 46), Scalar(77, 255, 255), mask);
		//imshow("mask", mask);
		bitwise_not(mask, mask);
		bitwise_and(frame, frame, dst, mask);
		imshow("front", dst);

	}
}

在这里插入图片描述

46. 直方图的反向投影

void Advance::back_project() {
	Mat src = imread("D:/images/hand.jpg");
	imshow("src", src);
	Mat sample = imread("D:/images/shand.png");
	imshow("sample", sample);
	
	Mat hsv;
	cvtColor(src, hsv, COLOR_BGR2HSV);
	
	int hbins = 48, s_bins = 48;
	int hist_size[] = { hbins, s_bins };
	int channels[] = { 0, 1 };
	Mat roi_hist;

	float h_ranges[] = { 0, 180 };
	float s_ranges[] = { 0, 255 };
	const float* ranges[] = { h_ranges, s_ranges };

	calcHist(&hsv, 1, channels, Mat(), roi_hist, 2, hist_size, ranges, true, false);
	normalize(roi_hist, roi_hist, 0, 255, NORM_MINMAX);
	MatND backproj;

	calcBackProject(&hsv, 1, channels, roi_hist, backproj, ranges, 1.0);
	imshow("back projection demo", backproj);

}

在这里插入图片描述

47. Harris角点检测

void Advance::harris_corner() {
	VideoCapture capture("D:/images/bike.avi");

	while (true) {
		if (!capture.isOpened()) {
			return;
		}

		int n = waitKey(50);
		if (n == 27) {
			break;
		}

		Mat frame;
		bool ret = capture.read(frame);
		if (!ret) {
			break;
		}

		Mat gray;
		cvtColor(frame, gray, COLOR_BGR2GRAY);
		Mat dst;
		double k = 0.04;
		int block_size = 2;
		int ksize = 3;
		cornerHarris(gray, dst, block_size, ksize, k);
		Mat dst_norm = Mat::zeros(frame.size(), frame.type());
		normalize(dst, dst_norm, 0, 255, NORM_MINMAX, -1, Mat());
		convertScaleAbs(dst_norm, dst_norm);

		RNG rng(143);
		for (int row = 0; row < frame.rows; ++row) {
			for (int col = 0; col < frame.cols; ++col) {
				int rsp = dst_norm.at<uchar>(row, col);
				if (rsp > 150) {
					int b = rng.uniform(0, 255);
					int g = rng.uniform(0, 255);
					int r = rng.uniform(0, 255);
					circle(frame, Point(col, row), 3, Scalar(b, g, r));
				}
			}
		}
		imshow("harris corner", frame);
	}
}

在这里插入图片描述

48. shitomas角点检测

void Advance::shitomas_corner_detector() {
	// shitomas脚点检测
	VideoCapture capture("D:/images/bike.avi");

	while (true) {
		if (!capture.isOpened()) {
			return;
		}
		int key = waitKey(50);
		if (key == 27) {
			break;
		}
		Mat frame;
		bool ret = capture.read(frame);
		if (!ret) {
			break;
		}
		Mat gray;
		cvtColor(frame, gray, COLOR_BGR2GRAY);
		vector<Point2f> corners;
		RNG rng(123);
		// qualty_level = 0.01代表response小于10的都过滤了
		double qualty_level = 0.01;
		goodFeaturesToTrack(gray, corners, 50, 0.01, 3, Mat());
		Mat shitomas = frame.clone();
		for (size_t i = 0; i < corners.size(); ++i) {
			int b = rng.uniform(0, 255);
			int g = rng.uniform(0, 255);
			int r = rng.uniform(0, 255);
			circle(frame, corners[i], 3, Scalar(b, g, r));
		}
		imshow("frame", frame);
		imshow("shitomas", shitomas);	   		 	  	  	
	}
}

在这里插入图片描述

49. 基于颜色对象的跟踪

本小节学习的重点：要进行颜色的提取，必须先转换到hsv的色彩空间上。

void Advance::color_trace() {
	VideoCapture capture("D:/images/blackboard.mp4");
	
	if (!capture.isOpened()) {
		return;
	}

	while (true)
	{
		int key = waitKey(50);
		if (key == 27) {
			break;
		}
		Mat frame;
		bool ret = capture.read(frame);
		if (!ret) {
			break;
		}
		Mat hsv, dst;
		cvtColor(frame, hsv, COLOR_BGR2HSV);
		// 0-10 43-255 46-255
		inRange(hsv, Scalar(0, 43, 46), Scalar(10, 255, 255), dst);
		imshow("frame", dst);
		// 形态学操作
		Mat morph;
		Mat se = getStructuringElement(MORPH_RECT, Size(8, 8));
		morphologyEx(dst, morph, MORPH_OPEN, se);
		imshow("morph", morph);
		// 查找轮廓contours
		vector<vector<Point>> contours;
		vector<Vec4i> hirearchy;
		findContours(morph, contours, hirearchy, RETR_EXTERNAL, CHAIN_APPROX_SIMPLE);

		double max_area = 0;
		size_t contour_index = -1;
		for (size_t t = 0; t < contours.size(); ++t) {
			double area = contourArea(contours[t]);
			if (area > max_area) {
				contour_index = t;
				max_area = area;
			}
		}
		printf("max area: %.1f\n", max_area);
		// 拟合
		if (contour_index >= 0) {
			RotatedRect rrt = minAreaRect(contours[contour_index]);
			ellipse(frame, rrt, Scalar(0, 255, 0), 2);
			circle(frame, rrt.center, 4, Scalar(255, 0, 0), 2);
		}
		imshow("color_trace", frame);
	}
}

在这里插入图片描述

50. 背景分析

void Advance::bg_extraction() {
	VideoCapture capture("D:/images/blackboard.mp4");

	if (!capture.isOpened()) {
		return;
	}
	auto pMOG2 = createBackgroundSubtractorMOG2(500, 1000, false);

	while (true)
	{
		int key = waitKey(50);
		if (key == 27) {
			break;
		}
		Mat frame;
		bool ret = capture.read(frame);
		if (!ret) {
			break;
		}
		Mat mask, bg_image;
		pMOG2->apply(frame, mask);
		pMOG2->getBackgroundImage(bg_image);

		Mat morph;
		Mat kernel = getStructuringElement(MORPH_RECT, Size(8, 8));
		morphologyEx(mask, morph, MORPH_OPEN, kernel);
		imshow("morph", morph);
		imshow("backgroud", bg_image);

	}
}

在这里插入图片描述

51. 光流法

void Advance::optical_analysis() {
	VideoCapture capture("D:/images/bike.avi");

	if (!capture.isOpened()) {
		return;
	}
	
	namedWindow("frame", WINDOW_AUTOSIZE);
	Mat old_frame, old_gray;
	capture.read(old_frame);
	cvtColor(old_frame, old_gray, COLOR_BGR2GRAY);
	vector<Point2f> feature_pts;
	double quality_level = 0.01;
	// shitomas角点检测
	goodFeaturesToTrack(old_gray, feature_pts, 200, quality_level, 3, Mat(), 3, false);

	vector<uchar> status;
	vector<float> err;
	Mat frame, gray;
	vector<Point2f> pts[2];
	pts[0].insert(pts[0].end(), feature_pts.begin(), feature_pts.end());
	TermCriteria criteria(TermCriteria::COUNT + TermCriteria::EPS, 10, 0.01);
	RNG rng(123);
	
	while (true)
	{
		int key = waitKey(50);
		if (key == 27) {
			break;
		}
		bool ret = capture.read(frame);
		if (!ret) {
			break;
		}

		capture.read(frame);
		cvtColor(frame, gray, COLOR_BGR2GRAY);
		// 光流分析
		calcOpticalFlowPyrLK(old_gray, gray, pts[0], pts[1], status, err, Size(21, 21), 3, criteria, 0);
		size_t i = 0, k = 0;
		for (i = 0; i < pts[1].size(); ++i) {
			if (status[i]) {
				pts[0][k] = pts[0][i];
				pts[1][k++] = pts[1][i];
				int b = rng.uniform(0, 255);
				int g = rng.uniform(0, 255);
				int r = rng.uniform(0, 255);
				circle(frame, pts[1][i], 2, Scalar(b, g, r), 2, 8, 0);
				line(frame, pts[0][i], pts[1][i], Scalar(b, g, r), 2, 8, 0);
			}
		}

		pts[0].resize(k);
		pts[1].resize(k);
		imshow("KLT-demo", frame);
		std::swap(pts[0], pts[1]);
		cv::swap(old_gray, gray);
	}
}

在这里插入图片描述

55. 基于均值迁移的视频分析

void Advance::mean_analysis() {
	VideoCapture capture("D:/images/blackboard.mp4");
	if (!capture.isOpened()) {
		printf("could not open the camera...\n");
	}
	namedWindow("frame", WINDOW_AUTOSIZE);
	Mat frame, hsv, hue, mask, hist, backproj;
	capture.read(frame);

	bool init = false;
	Rect trackWindow;

	int hsize = 16;
	float hranges[] = { 0,180 };
	const float* ranges = hranges;
	
	Rect selection = selectROI("MeanShift Demo", frame, true, false);
	while (true) {
		bool ret = capture.read(frame);
		if (!ret) {
			break;
		}
		cvtColor(frame, hsv, COLOR_BGR2HSV);
		inRange(hsv, Scalar(156, 43, 46), Scalar(180, 255, 255), mask);

		int ch[] = { 0, 0 };
		hue.create(hsv.size(), hsv.depth());
		mixChannels(&hsv, 1, &hue, 1, ch, 1);


		if (init) {
			Mat roi(hue, selection), maskroi(mask, selection);
			calcHist(&roi, 1, 0, maskroi, hist, 1, &hsize, &ranges);
			normalize(hist, hist, 0, 255, NORM_MINMAX);
			trackWindow = selection;
			init = false;
		}
		// ms
		calcBackProject(&hue, 1, 0, hist, backproj, &ranges);
		backproj &= mask;
		meanShift(backproj, trackWindow, TermCriteria(TermCriteria::COUNT | TermCriteria::EPS, 10, 1));
		rectangle(frame, trackWindow, Scalar(0, 0, 255), 3, LINE_AA);
		imshow("MeanShift Demo", frame);
		char c = waitKey(1);
		if (c == 27) {
			break;
		}
	}

}

相关阅读:
阿里云刘洋：基于eBPF的Kubernetes可观测最佳实践
 VSCode C++环境配置及测试运行
 jvm server和client模式切换
 2022年mvnrepository跳过人机验证
 I/O多路转接
 【车间调度】基于模拟退火优化算法的的并行车间机器优化调度（Matlab代码实现）
[附源码]SSM计算机毕业设计血库管理系统JAVA
【深度学习】特征融合的重要方法 | 张量的拼接 | torch.cat()函数 | torch.add(函数
 MobPush HarmonyOS NEXT 版本集成指南
 Anchor-free目标检测综述 -- Dense Prediction篇
原文地址：https://blog.csdn.net/HELLOWORLD2424/article/details/126540154

OpenCV4 图像处理与视频分析实战教程 笔记

14. 图像直方图的比较

15. 使用LUT

19. 自定义滤波

20. 图像的梯度

21. 拉普拉斯算子和图像的锐化

22. USM锐化

23. 图像噪声

24. 图像去噪

25. 边缘保留滤波

26. canny边缘提取加上trackbar演示

27. 二值图像的概念

28. 全局阈值

29. 自适应阈值

30. 联通件的扫描

31. 带统计信息联通件扫描

32. 轮廓的发现和绘制

33. 图像轮廓的计算

34. 轮廓匹配

3.5 轮廓逼近和拟合

36 霍夫直线检测

37 霍夫直线检测2

38. 霍夫圆检测

39. 图像形态学操作

40. 图像的开闭

41. 形态学梯度

42. 更多形态学操作

43. 综合练习 找出卫星图岛屿轮廓

44. 视频读写api

45. 色彩空间转换

46. 直方图的反向投影

47. Harris角点检测

48. shitomas角点检测

49. 基于颜色对象的跟踪

50. 背景分析

51. 光流法

55. 基于均值迁移的视频分析

OpenCV4 图像处理与视频分析实战教程笔记

43. 综合练习找出卫星图岛屿轮廓