目录

1、积分图像

2、图像分割--漫水填充

3、图像分割--分水岭法

4、Harris角点检测

---

1、积分图像

 

//积分图像
int test1()
{
	//创建一个16×16全为1的矩阵,因为256=16×16
	Mat img = Mat::ones(16, 16, CV_32FC1);
	//在图像中加入随机噪声
	RNG rng(10086);
	for (int y = 0; y < img.rows; y++)
	{
		for (int x = 0; x < img.cols; x++)
		{
			float d = rng.uniform(-0.5, 0.5);
			img.at<float>(y, x) = img.at<float>(y, x) + d;
		}
	}
	//计算标准求和积分
	Mat sum;
	integral(img, sum);
	//为了便于显示,转成CV_8U格式
	Mat sum8U = Mat_<uchar>(sum);
	namedWindow("sum8U", WINDOW_NORMAL);
	imshow("sum8U", sum8U);

	//计算平方求和积分
	Mat sqsum;
	integral(img, sum, sqsum);//为了便于显示,转成CV_8U格式
	Mat sqsum8U = Mat_<uchar>(sqsum);
	namedWindow("sqsum8U", WINDOW_NORMAL);
	imshow("sqsum8U", sqsum8U);

	//计算倾斜求和积分
	Mat tilted;
	integral(img, sum, sqsum, tilted);//为了便于显示,转成CV_8U格式
	Mat tilted8U = Mat_<uchar>(tilted);
	namedWindow("tilted8U", WINDOW_NORMAL);
	imshow("tilted8U", tilted8U);

	waitKey(0);
	return 0;
}

2、图像分割--漫水填充

 

//图像分割--漫水填充
int test2()
{
	system("color 02");//将DOS界面调成白底黑字
	Mat img = imread("F:/testMap/lena.png");
	if (!(img.data))
	{
		cout << "读取图像错误,请确认图像文件是否正确" << endl;
		return -1;
	}
	RNG rng(10086);//随机数,用于随机生成像素

	//设置操作标志flags
	int connectivity = 4;//连通邻域方式
	int maskVal = 255;//掩码图像的数值
	int flags = connectivity | (maskVal << 8) | FLOODFILL_FIXED_RANGE;//漫水填充操作方式标志

	//设置与选中像素点的差值
	Scalar loDiff = Scalar(20, 20, 20);
	Scalar upDiff = Scalar(20, 20, 20);

	//声明掩模矩阵变量
	Mat mask = Mat::zeros(img.rows + 2, img.cols + 2, CV_8UC1);

	while (true)
	{
		//随机产生图像中某一像素点
		int py = rng.uniform(0, img.rows - 1);
		int px = rng.uniform(0, img.cols - 1);
		Point point = Point(px, py);

		//彩色图像中填充的像素值
		Scalar newVal = Scalar(rng.uniform(0, 255), rng.uniform(0, 255), rng.uniform(0, 255));

		//漫水填充函数
		int area = floodFill(img, mask, point, newVal, &Rect(), loDiff, upDiff, flags);

		//输出像素点和填充的像素数目
		cout << "像素点x: " << point.x << " y : " << point.y << "填充像数数目:" << area << endl;

		//输出填充的图像结果
		imshow("填充的彩色图像", img);
		imshow("掩模图像", mask);

		//判断是否结束程序
		int c = waitKey(0);
		if ((c & 255) == 27)
		{
			break;
		}
	}
	waitKey(0);
	return 0;
}

3、图像分割--分水岭法

 

 

//图像分割--分水岭法
int test3()
{
	Mat img, imgGray, imgMask, img_;
	Mat maskWaterShed; //watershed()函数的参数

	img = imread("F:/testMap/lenaw.png"); //含有标记的图像
	img_ = imread("F:/testMap/lena.png"); //原图像
	cvtColor(img, imgGray, COLOR_BGR2GRAY);
	
	//二值化并开运算
	threshold(imgGray, imgMask, 230, 255, THRESH_BINARY);
	Mat k = getStructuringElement(0, Size(3, 3));
	morphologyEx(imgMask, imgMask, MORPH_OPEN, k);
	imshow("含有标记的图像", img);
	imshow("原图像", img_);

	vector<vector<Point>> contours;
	vector<Vec4i> hierarchy;
	findContours(imgMask, contours, hierarchy, RETR_CCOMP, CHAIN_APPROX_SIMPLE);

	//在maskWaterShed上绘制轮廓,用于输入分水岭算法
	maskWaterShed = Mat::zeros(imgMask.size(),CV_32S); 
	for (int index = 0; index < contours.size(); index++)
	{
		drawContours(maskWaterShed, contours, index, Scalar::all(index + 1), -1, 8, hierarchy, INT_MAX);
	}
	//分水岭算法需要对原图像进行处理
	watershed(img_, maskWaterShed);
	
	vector<Vec3b> colors;// 随机生成几种颜色
	for (int i = 0; i < contours.size(); i++)
	{
		int b = theRNG().uniform(0, 255); 
		int g = theRNG().uniform(0, 255); 
		int r = theRNG().uniform(0, 255);
		colors.push_back(Vec3b((uchar)b, (uchar)g, (uchar)r));
	}

	Mat resultImg = Mat(img.size(), CV_8UC3); // 显示图像
	for(int i = 0; i < imgMask.rows; i++)
	{
		for (int j = 0; j < imgMask.cols; j++)
		{
			//绘制每个区域的颜色
			int index = maskWaterShed.at<int>(i, j);
			if (index == -1)//区域间的值被置为 - 1(边界)
			{
				resultImg.at<Vec3b>(i, j) = Vec3b(255, 255, 255);
			}
			else if (index <= 0 || index > contours.size())//没有标记清楚的区域被置为0
			{
				resultImg.at<Vec3b>(i, j) = Vec3b(0, 0, 0);
			}
			else//其他每个区域的值保持不变: 1,2,…,contours.size()
			{
				resultImg.at<Vec3b>(i, j) = colors[index - 1]; //把些区域绘制成不同颜色
			}
		}
	}
	imshow("resultImg", resultImg);
	resultImg = resultImg * 0.8 + img_*0.2;
	//addWeighted(resultImg,0.8,img_, 0.2,0, resultImg); 
	imshow("分水岭结果", resultImg);

	//绘制每个区域的图像
	for (int n = 1; n <= contours.size(); n++)
	{
		Mat resImagel = Mat(img.size(), CV_8UC3);//声明一个最后要显示的图像
		for (int i = 0; i < imgMask.rows; i++)
		{
			for (int j = 0; j < imgMask.cols; j++)
			{
				int index = maskWaterShed.at<int>(i, j);
				if (index == n)
					resImagel.at<Vec3b>(i, j) = img_.at<Vec3b>(i, j);
				else
					resImagel.at<Vec3b>(i, j) = Vec3b(0, 0, 0);
			}
		}
		//显示图像
		imshow(to_string(n), resImagel);
	}

	waitKey(0);
	return 0;
}

4、Harris角点检测

 

 

 

 

//Harris角点检测
int test4()
{
	Mat img = imread("F:/testMap/lena.png", IMREAD_COLOR);
	if (!img.data)
	{
		cout << "读取图像错误,请确认图像文件是否正确" << endl;
		return -1;
	}
	//转成灰度图像
	Mat gray;
	cvtColor(img, gray,COLOR_BGR2GRAY);
	
	//计算Harris系数
	Mat harris;
	int blockSize = 2; //邻域半径
	int apertureSize = 3;
	cornerHarris(gray,harris, blockSize,apertureSize,0.04);

	//归一化便于进行数值比较和结果显示
	Mat harrisn;
	normalize(harris, harrisn, 0, 255, NORM_MINMAX);//将图像的数据类型变成CV_8U
	convertScaleAbs(harrisn, harrisn);

	//寻找Harris角点
	vector<KeyPoint> keyPoints;
	for (int row = 0; row < harrisn.rows; row++)
	{
		for (int col = 0; col < harrisn.cols; col++)
		{
			int R = harrisn.at<uchar>(row, col);
			if (R >125)
			{
				//向角点存入KeyPoint中
				KeyPoint keyPoint;
				keyPoint.pt.y =row;
				keyPoint.pt.x = col;
				keyPoints.push_back(keyPoint);
			}
		}
	}
	//绘制角点与显示结果
	drawKeypoints(img,keyPoints,img); 
	imshow("系数矩阵", harrisn);
	imshow("Harris角点", img) ;

	waitKey(0);
	return 0;
}