数字图像处理实训-基于C++实现的图像连通域提取与形状特征分析源码+代码注释.zip

// ShapeFeatures.cpp : 此文件包含 "main" 函数。程序执行将在此处开始并结束。 // #include <iostream> #include <opencv2/core/core.hpp> #include <opencv2/highgui/highgui.hpp> #include <opencv2/opencv.hpp> #include <cstring> #include <vector> using namespace std; using namespace cv; typedef struct{ long int cir; //周长 long int area; //面积 double R;//圆心度 Point massPt;//重心 vector<Point> inPt; //内部点 vector<Point> edgePt;//边缘点 }Shape; //图形类型 /////////////////////////////////////////////////辅助函数////////////////////////////////////////////////////////////// //彩色图像灰度化 Mat Color2Gray(Mat const& SrcImg) { int rows = SrcImg.rows; int cols = SrcImg.cols; Mat mGray; mGray.create(rows, cols, CV_8UC1); int nMaxValue; int bValue, gValue, rValue; for (int i = 0; i < rows; i++) { for (int j = 0; j < cols; j++) { bValue = SrcImg.at<Vec3b>(i, j)[0]; gValue = SrcImg.at<Vec3b>(i, j)[1]; rValue = SrcImg.at<Vec3b>(i, j)[2]; nMaxValue = bValue; if (nMaxValue < gValue) { nMaxValue = gValue; } if (nMaxValue < rValue) { nMaxValue = rValue; } mGray.at<uchar>(i, j) = saturate_cast<uchar>(nMaxValue); } } return mGray; } //图像二值化(状态法) Mat StateMethord(const Mat& srcImg) { Mat gray_img, dst; int cn = srcImg.channels();//通道数 if (cn == 3) gray_img = Color2Gray(srcImg); else gray_img = srcImg.clone(); int height = gray_img.rows;//行数 int width = gray_img.cols;//列数 dst.create(gray_img.size(), gray_img.type()); //峰谷法获取阙值- int nThreshold = 0, nNewThreshold = 0; int hist[256]; memset(hist, 0, sizeof(hist));//定义灰度统计数组 int lS1, lS2;//类1,2像素数 double lP1, lP2;//类1,2质量矩 double meanvalue1, meanvalue2;//类1,2灰度均值 int IterationTimes;//迭代次数 int graymax = 255, graymin = 0;//灰度最大最小值 int i, j, k; for (i = 0; i < height; i++)//统计出各个灰度值的像素数 for (j = 0; j < width; j++) { int gray = gray_img.at<uchar>(i, j); hist[gray]++; } //给阙值赋迭代初值 nNewThreshold = int((graymax + graymin) / 2); //迭代求最佳阙值 for (IterationTimes = 0; nThreshold != nNewThreshold && IterationTimes < 100; IterationTimes++) { nThreshold = nNewThreshold; lP1 = lP2 = 0.0; lS1 = lS2 = 0; //求类1的质量矩和像素总数 for (k = graymin; k <= nThreshold; k++) { lP1 += double(k) * hist[k]; lS1 += hist[k]; } //计算类1均值 meanvalue1 = lP1 / lS1; //求类2的质量矩和像素总数 for (k = nThreshold + 1; k <= graymax; k++) { lP2 += double(k) * hist[k]; lS2 += hist[k]; } //计算类2均值 meanvalue2 = lP2 / lS2; //获取新的阙值 nNewThreshold = int((meanvalue1 + meanvalue2) / 2); } //二值化图像 for (i = 0; i < height; i++) for (j = 0; j < width; j++) { if (gray_img.at<uchar>(i, j) < nNewThreshold) dst.at<uchar>(i, j) = 0; else dst.at<uchar>(i, j) = 255; } return dst; } //图像二值化(判断分析法) Mat OSTU(const Mat& srcImg) { //创建灰度图 Mat gray_img, dst; int cn = srcImg.channels();//通道数 if (cn == 3) gray_img= Color2Gray(srcImg); else gray_img = srcImg.clone(); int height = gray_img.rows;//行数 int width = gray_img.cols;//列数 dst.create(gray_img.size(), gray_img.type()); //判断分析法（Ostu）获取阙值 int nThreshold = 0; int hist[256]; memset(hist, 0, sizeof(hist));//定义灰度统计数组 int lS, lS1, lS2;//总像素数，类1,2像素数 double lP, lP1;//总质量矩，类1质量矩 double meanvalue1, meanvalue2;//类1,2灰度均值 int IterationTimes;//迭代次数 double Dis1, Dis2, CinDis, CoutDis, max;//方差，类1，类2，组内，组间 int graymax = 255, graymin = 0;//灰度最大最小值 int i, j, k; //统计出各个灰度值的像素数 for (i = 0; i < height; i++) for (j = 0; j < width; j++) { int gray = gray_img.at<uchar>(i, j); hist[gray]++; } lP = 0.0; lS = 0; if (graymin == 0) graymin++; //计算总的质量矩lP和像素总数lS for (k = graymin; k <= graymax; k++) { lP += double(k) * double(hist[k]); lS += hist[k]; } max = 0.0; lS1 = lS2 = 0; lP1 = 0.0; Dis1 = Dis2 = CinDis = CoutDis = 0.0; double ratio = 0.0; //求阙值 for (k = graymin; k < graymax; k++) { lS1 += hist[k]; if (!lS1) { continue; } lS2 = lS - lS1; if (lS2 == 0) { break; } lP1 += double(k) * double(hist[k]); meanvalue1 = lP1 / lS1; for (int n = graymin; n <= k; n++) Dis1 += double((n - meanvalue1) * (n - meanvalue1) * hist[n]); meanvalue2 = (lP - lP1) / lS2; for (int m = k + 1; m < graymax; m++) Dis2 += double((m - meanvalue2) * (m - meanvalue2) * hist[m]); CinDis = Dis1 + Dis2; CoutDis = double(lS1) * double(lS2) * (meanvalue1 - meanvalue2) * (meanvalue1 - meanvalue2); if (CinDis != 0) ratio = CoutDis / CinDis; if (ratio > max) { max = ratio; nThreshold = k; } } //二值化图像 for (i = 0; i < height; i++) for (j = 0; j < width; j++) { if (gray_img.at<uchar>(i, j) < nThreshold) dst.at<uchar>(i, j) = 0; else dst.at<uchar>(i, j) = 255; } return dst; } //滑动条回调函数 void on_ThreshChange(int, void*)//回调函数 { } //找到图形中面积的最值 void findM(vector<Shape>& shapes, long int& areaMax, int& areaMin) { int max = 0; int min = 99999999; for (vector<Shape>::iterator iter = shapes.begin() + 1; iter != shapes.end(); iter++) { if (min > iter->area) min = iter->area; if (max < iter->area) max = iter->area; } areaMax = max; areaMin = min; } //读取二进制文件 Mat readbyte(const char* path) { FILE* fp; if (fopen_s(&fp, path, "rb") != 0) { printf("cannot open file for read\n"); waitKey();//opencv中常用waitKey(n),n<=0表示一直 exit(0); } int cols = 2076; int rows = 2816; int bands = 3;// 波段数 int pixels = cols * rows * bands; unsigned char* data = new uchar[pixels * sizeof(uchar)];//uchar等效于char，此处可以借鉴这样的预分配内存方法 fread(data, sizeof(uchar), pixels, fp);//fread函数的用法 fclose(fp); Mat M(rows, cols, CV_8UC3, Scalar(0, 0, 0)); unsigned char* ptr = M.data;//data是存数据的部分指针 for (int i = 0; i < rows; i++) { for (int j = 0; j < cols; j++) { for (int k = 0; k < 3; k++) { ptr[(i * cols + j) * 3 + k] = data[(i * cols + j) * 3 + k];//BIP } } } return M; } /////////////////////////////////////////////////核心函数//////////////////////////////////////////////////////////// //①连通成分标记 //种子填充法 int Cc_FeedFill(const Mat& binImg, Mat& label) { //如果为空或并非单通道图像，则返回-1 if (binImg.empty() || binImg.type() != CV_8UC1)return -1; int labels = 0; int cols = binImg.cols; int rows = binImg.rows; label = Mat::zeros(rows, cols, CV_8UC1); vector<int>vt; for (int i = 0; i < rows; i++) { for (int j = 0; j < cols; j++) { if (binImg.at<uchar>(i * cols + j) == 255 && label.at<uchar>(i * cols + j) == 0) { vt.push_back(i * cols + j); labels++; while (vt.size() != 0) { int px = *(vt.end() - 1); vt.pop_back(); label.at<uchar>(px) = labels; //查询上下右左是否为连通分量 if (px > cols && binImg.at<uchar>(px - cols) == 255 && label.at<uchar>(px - cols) == 0) vt.push_back(px - cols); if (px < (rows - 1) * cols && binImg.at<uchar>(px + cols) == 255 && label.at<uchar>(px + cols) == 0) vt.push_back(px + cols); if ((px + 1) % cols != 0 && binImg.at<uchar>(px + 1) == 255 && label.at<uchar>(px + 1) == 0) vt.push_back(px + 1); if ((px) % cols != 0 && binImg.at<uchar>(px - 1) == 255 && label.at<uchar>(px -