C++ opencv基于OTSU图像多阈值分割

vector<double> sub(vector<int> A, vector<double> B) //减法
{vector<double> C;C.resize(3);for (int i = 0; i < A.size(); i++){C[i] = A[i] - B[i];}return C;
}vector<double> multi(vector<double> A, int b)//乘法
{vector<double> C;C.resize(3);for (int i = 0; i < A.size(); i++)   {C[i] = A[i] * b;}return C;
}vector<double> multi(vector<double> A, vector<double> B)//乘法
{vector<double> C;C.resize(3);for (int i = 0; i < A.size(); i++){C[i] = A[i] * B[i];}return C;
}vector<double> divi(vector<double> A, vector<double> B)//处罚
{vector<double> C;C.resize(3);for (int i = 0; i < A.size(); i++){if(B[i]==0){continue;}C[i] = A[i] / B[i];}return C;
}/*
* 实现python的linspace函数功能
*/
std::vector<double> linspace(int start_in, int end_in, int num_in)
{std::vector<double> linspaced;double start = static_cast<double>(start_in);double end = static_cast<double>(end_in);double num = static_cast<double>(num_in);if (num == 0) { return linspaced; }if (num == 1){linspaced.push_back(start);return linspaced;}double delta = (end - start) / (num - 1);for (int i = 0; i < num - 1; ++i){linspaced.push_back(start + delta * i);}linspaced.push_back(end); return linspaced;
}
/*
* vector容器第n个到m个元素求和
*/
double compute_sum(vector<double>& Vec, int start, int end)
{double total = 0;for (int i = start; i < end; i++){total += Vec[i];}  return total;
}/*
* vector容器dot点乘
*/
double compute_dot(vector<double>& Vec1, vector<double>& Vec2, int start, int end)
{double total = 0;double temp = 0;for (int i = start; i < end; i++){temp = Vec1[i]*Vec2[i];total = total + temp;}return total;
}void otsu3layers(Mat &input)
{//Mat img = input.clone();int high = input.rows;int wide = input.cols;Mat hist;const int bins[1] = { 256 };float hranges[2] = { 0,256 };const float* ranges[1] = { hranges };calcHist(&input, 1, 0, Mat(), hist, 1, bins, ranges);//hist256*1，每个灰度级中所含的像素数vector<double>p;//每个像素所占总像素数的比例std::vector <double>lsp;double temp = 0;for (int i = 0; i < hist.rows; i++){for (int j = 0; j < hist.cols; j++){temp = hist.at<float>(i, j) / (high * wide);p.push_back(temp);}}lsp = linspace(1, 256, 256);//.reshape没写double maxvar = 0;int th1 = 0;int th2 = 0;int cnt = 0;double w0 = 0;double w1 = 0;double w2 = 0;double u0 = 0;double u1 = 0;double u2 = 0;double varbetween = 0;vector<double>W;vector<int>U;vector<double>tempp;for (int k1 = 1; k1 < 257; k1++){for (int k2 = k1 + 1; k2 < 257; k2++){w0 = compute_sum(p, 0, k1);//0.00118w1 = compute_sum(p, k1, k2);//0.00076w2 = compute_sum(p, k2, 256);//0.998u0 = compute_dot(lsp, p, 0, k1);//0.00118u1 = compute_dot(lsp, p, k1, k2);//0.00152u2 = compute_dot(lsp, p, k2, 256);//53.78W.push_back(w0);W.push_back(w1);W.push_back(w2);U.push_back((int)u0);U.push_back((int)u1);U.push_back((int)u2);int UT = accumulate(U.begin(), U.end(), 0);//53tempp = multi(W, UT);tempp = sub(U, tempp);tempp = multi(tempp, tempp);tempp = divi(tempp, W);varbetween = compute_sum(tempp, 0, 3);if (varbetween > maxvar){cnt = cnt + 1;maxvar = varbetween;th1 = k1;th2 = k2;}W.clear();U.clear();tempp.clear();/*cout << "阈值为" << th1 << "和" << th2 << endl;*/}}for (int row = 0; row < high; row++){uchar* cur = input.ptr<uchar>(row);for (int col = 0; col < wide; col++){uchar gray = cur[col];/*if (gray < th1){gray = 0;cur[col] = gray;}else if (gray > th1 && gray < th2){gray = 127;cur[col] = gray;}else{gray = 255;cur[col] = gray;}*/if (gray < th2){gray = 255;cur[col] = gray;}else{gray = 0;cur[col] = gray;}}}}

参考自：https://www.cnblogs.com/Linkdom/p/16122676.html#%E5%9F%BA%E4%BA%8Eotsu%E7%9A%84%E5%9B%BE%E5%83%8F%E5%A4%9A%E9%98%88%E5%80%BC%E5%88%86%E5%89%B2%E7%AE%97%E6%B3%95

C++ opencv基于OTSU图像多阈值分割相关推荐

c++opencv汉字分割_机器学习小白，还不快pick一下——【视觉与图像：阈值分割】...
" 前言:安利Python来开发OpenCV的原因其实细心的小伙伴早在?[视觉与图像]Python+OpenCV教程入门篇就找到了想要的答案.(点蓝字即可打开) " 今天周五了! ...
图像的阈值分割(Optimum Thresholding)
摘要:本实验主要实现最优阈值.图像分割就是把图像中具有特殊涵义的不同区域分开来,这些区域是互不相交的,每个区域都满足特定区域的一致性.图像分割是图像处理中的重要问题,也是计算机视觉研究中的一个 ...
红外目标图像中阈值分割方法的比较与研究
红外目标图像中阈值分割方法的比较与研究摘要:本文主要以红外图像目标检测技术为背景,在图像阈值分割中以最大熵准则及遗传算法为基础,研究了一维最大熵值法(KSW法)及基于遗传算法的KSW熵法单阈值. ...
Python对DICOM图像进行阈值分割
Python对Dicom图像进行处理,离不开pydicom,opencv-python,matplotlib,numpy四个代码库,安装完成这四个代码库后, 可以读取Dicom图像,并对图像进行处理, ...
图像二值化方法及适用场景分析（OTSU Trangle 自适应阈值分割）
图像二值化应用场景二值图像定义阈值获取的方法手动阈值法自动阈值法灰度均值法基于直方图均值法 OTSU Triangle 自适应均值阈值分割方法总结参考文献应用场景二值图像处理与分 ...
基于OpenCV和C++实现最大阈值分割算法
代码如下:: /******************************************************************************************** ...
matlab中图像的阈值分割,基于MATLAB的图像阈值分割技术汇总
数字图像处理课程论文基于MATLAB的图像阈值分割技术摘要:本文主要针对图像阈值分割做一个基于MATLAB的分析.通过双峰法,迭代法以及OUTS法三种算法来实现图像阈值分割,并且就这三种算法做了 ...
python+OpenCV图像处理（五）图像的阈值分割
图像的阈值处理一幅图像包括目标物体.背景还有噪声,要想从多值的数字图像中直接提取出目标物体,常用的方法就是设定一个阈值T,用T将图像的数据分成两部分:大于T的像素群和小于T的像素群.这是研究灰度变换 ...
opencv学习笔记10：阈值分割
阈值分割像素图原始图像像素图见下面红色线:标注一条阈值线二进制阈值化首先设定一条阀值线如127 大于127的像素点灰度值设为最大(如unit8的格式为255) 小于127的像素点灰度值设 ...

C++ opencv基于OTSU图像多阈值分割

C++ opencv基于OTSU图像多阈值分割相关推荐

最新文章

热门文章