时间:2022-12-01 10:56:35 | 栏目:C代码 | 点击:次
之前有提到给灰度图上色的需求,在此基础上,还有一种需求,就是基于另一张参考灰度图的色板来给当前的灰度图上色,比如参考灰度图的数值区间为-10到10,颜色从蓝到绿再到红,而当前的灰度图的数据区间为-1到1,若基于参考灰度图的色板确定数据对应的颜色,则当前灰度图的颜色应该在绿色左右波动。
下方为具体实现函数和测试代码。
/** * @brief GrayToColorFromOther 灰度图上色,基于参考灰度图的色板 * @param phase1 输入的灰色图像,通道为1,提供色板 * @param phase2 输入的灰色图像,通道为1,基于phase1的色板绘色 * @return 上色后的图像 */ cv::Mat GrayToColorFromOther(cv::Mat &phase1, cv::Mat &phase2) { CV_Assert(phase1.channels() == 1); CV_Assert(phase2.channels() == 1); if (phase1.empty() || phase2.empty()) { cv::Mat result = cv::Mat::zeros(100, 100, CV_8UC3); return result; } cv::Mat temp, result, mask; double max1, min1; int row = phase2.rows; int col = phase2.cols; // 确定参考灰度图的数据范围 cv::minMaxIdx(phase1, &min1, &max1, nullptr, nullptr, phase1 == phase1); // 将当前灰度图以参考灰度图的数据范围作标准,进行数据变换 temp = phase2.clone(); for (int i = 0; i < row; ++i) { float *t2 = temp.ptr<float>(i); for (int j = 0; j < col; ++j) { t2[j] = 255.0f*(phase2.at<float>(i, j) - min1) / (max1 - min1); } } temp.convertTo(temp, CV_8UC1); // 创建掩膜,目的是为了隔离nan值的干扰 mask = cv::Mat::zeros(phase2.size(), CV_8UC1); mask.setTo(255, phase2 == phase2); // 初始化三通道颜色图 cv::Mat color1, color2, color3; color1 = cv::Mat::zeros(temp.size(), temp.type()); color2 = cv::Mat::zeros(temp.size(), temp.type()); color3 = cv::Mat::zeros(temp.size(), temp.type()); // 基于灰度图的灰度层级,给其上色,最底的灰度值0为蓝色(255,0,0),最高的灰度值255为红色(0,0,255),中间的灰度值127为绿色(0,255,0) for (int i = 0; i < row; ++i) { uchar *c1 = color1.ptr<uchar>(i); uchar *c2 = color2.ptr<uchar>(i); uchar *c3 = color3.ptr<uchar>(i); uchar *r = temp.ptr<uchar>(i); uchar *m = mask.ptr<uchar>(i); for (int j = 0; j < col; ++j) { if (m[j] == 255) { if (r[j] > (3 * 255 / 4) && r[j] <= 255) { c1[j] = 255; c2[j] = 4 * (255 - r[j]); c3[j] = 0; } else if (r[j] <= (3 * 255 / 4) && r[j] > (255 / 2)) { c1[j] = 255 - 4 * (3 * 255 / 4 - r[j]); c2[j] = 255; c3[j] = 0; } else if (r[j] <= (255 / 2) && r[j] > (255 / 4)) { c1[j] = 0; c2[j] = 255; c3[j] = 4 * (255 / 2 - r[j]); } else if (r[j] <= (255 / 4) && r[j] >= 0) { c1[j] = 0; c2[j] = 255 - 4 * (255 / 4 - r[j]); c3[j] = 255; } else { c1[j] = 0; c2[j] = 0; c3[j] = 0; } } } } // 三通道合并,得到颜色图 vector<cv::Mat> images; images.push_back(color3); images.push_back(color2); images.push_back(color1); cv::merge(images, result); return result; }
#include<iostream> #include<opencv2/opencv.hpp> #include<ctime> using namespace std; using namespace cv; void UnitPolar(int squaresize, cv::Mat& mag,cv::Mat& ang); void UnitCart(int squaresize, cv::Mat& x, cv::Mat& y); cv::Mat GrayToColor(cv::Mat &phase); cv::Mat GrayToColorFromOther(cv::Mat &phase1, cv::Mat &phase2); int main(void) { cv::Mat mag, ang,result,result2; UnitPolar(2001, mag, ang); mag.at<float>(10, 10) = nan(""); cv::Mat mag2 = mag / 2; result = GrayToColor(mag); result2= GrayToColorFromOther(mag,mag2); system("pause"); return 0; } void UnitPolar(int squaresize, cv::Mat& mag,cv::Mat& ang) { cv::Mat x; cv::Mat y; UnitCart(squaresize, x, y); //产生指定范围内的指定数量点数,相邻数据跨度相同 // OpenCV自带的转换有精度限制,导致结果有一定差异性 //cv::cartToPolar(x, y, mag, ang, false); //坐标转换 mag = cv::Mat(x.size(), x.type()); ang = cv::Mat(x.size(), x.type()); int row = mag.rows; int col = mag.cols; float *m, *a, *xx, *yy; for (int i = 0; i < row; ++i) { m = mag.ptr<float>(i); a = ang.ptr<float>(i); xx = x.ptr<float>(i); yy = y.ptr<float>(i); for (int j = 0; j < col; ++j) { m[j] = sqrt(xx[j] * xx[j] + yy[j] * yy[j]); a[j] = atan2(yy[j], xx[j]); } } } void UnitCart(int squaresize, cv::Mat& x, cv::Mat& y) { CV_Assert(squaresize % 2 == 1); x.create(squaresize, squaresize, CV_32FC1); y.create(squaresize, squaresize, CV_32FC1); //设置边界 x.col(0).setTo(-1.0); x.col(squaresize - 1).setTo(1.0f); y.row(0).setTo(1.0); y.row(squaresize - 1).setTo(-1.0f); float delta = 2.0f / (squaresize - 1.0f); //两个元素的间隔 //计算其他位置的值 for (int i = 1; i < squaresize - 1; ++i) { x.col(i) = -1.0f + i * delta; y.row(i) = 1.0f - i * delta; } } /** * @brief GrayToColor 灰度图上色 * @param phase 输入的灰色图像,通道为1 * @return 上色后的图像 */ cv::Mat GrayToColor(cv::Mat &phase) { CV_Assert(phase.channels() == 1); cv::Mat temp, result, mask; // 将灰度图重新归一化至0-255 cv::normalize(phase, temp, 255, 0, cv::NORM_MINMAX); temp.convertTo(temp, CV_8UC1); // 创建掩膜,目的是为了隔离nan值的干扰 mask = cv::Mat::zeros(phase.size(), CV_8UC1); mask.setTo(255, phase == phase); // 初始化三通道颜色图 cv::Mat color1, color2, color3; color1 = cv::Mat::zeros(temp.size(), temp.type()); color2 = cv::Mat::zeros(temp.size(), temp.type()); color3 = cv::Mat::zeros(temp.size(), temp.type()); int row = phase.rows; int col = phase.cols; // 基于灰度图的灰度层级,给其上色,最底的灰度值0为蓝色(255,0,0),最高的灰度值255为红色(0,0,255),中间的灰度值127为绿色(0,255,0) // 不要惊讶蓝色为什么是(255,0,0),因为OpenCV中是BGR而不是RGB for (int i = 0; i < row; ++i) { uchar *c1 = color1.ptr<uchar>(i); uchar *c2 = color2.ptr<uchar>(i); uchar *c3 = color3.ptr<uchar>(i); uchar *r = temp.ptr<uchar>(i); uchar *m = mask.ptr<uchar>(i); for (int j = 0; j < col; ++j) { if (m[j] == 255) { if (r[j] > (3 * 255 / 4) && r[j] <= 255) { c1[j] = 255; c2[j] = 4 * (255 - r[j]); c3[j] = 0; } else if (r[j] <= (3 * 255 / 4) && r[j] > (255 / 2)) { c1[j] = 255 - 4 * (3 * 255 / 4 - r[j]); c2[j] = 255; c3[j] = 0; } else if (r[j] <= (255 / 2) && r[j] > (255 / 4)) { c1[j] = 0; c2[j] = 255; c3[j] = 4 * (255 / 2 - r[j]); } else if (r[j] <= (255 / 4) && r[j] >= 0) { c1[j] = 0; c2[j] = 255 - 4 * (255 / 4 - r[j]); c3[j] = 255; } else { c1[j] = 0; c2[j] = 0; c3[j] = 0; } } } } // 三通道合并,得到颜色图 vector<cv::Mat> images; images.push_back(color3); images.push_back(color2); images.push_back(color1); cv::merge(images, result); return result; } /** * @brief GrayToColorFromOther 灰度图上色,基于参考灰度图的色板 * @param phase1 输入的灰色图像,通道为1,提供色板 * @param phase2 输入的灰色图像,通道为1,基于phase1的色板绘色 * @return 上色后的图像 */ cv::Mat GrayToColorFromOther(cv::Mat &phase1, cv::Mat &phase2) { CV_Assert(phase1.channels() == 1); CV_Assert(phase2.channels() == 1); if (phase1.empty() || phase2.empty()) { cv::Mat result = cv::Mat::zeros(100, 100, CV_8UC3); return result; } cv::Mat temp, result, mask; double max1, min1; int row = phase2.rows; int col = phase2.cols; // 确定参考灰度图的数据范围 cv::minMaxIdx(phase1, &min1, &max1, nullptr, nullptr, phase1 == phase1); // 将当前灰度图以参考灰度图的数据范围作标准,进行数据变换 temp = phase2.clone(); for (int i = 0; i < row; ++i) { float *t2 = temp.ptr<float>(i); for (int j = 0; j < col; ++j) { t2[j] = 255.0f*(phase2.at<float>(i, j) - min1) / (max1 - min1); } } temp.convertTo(temp, CV_8UC1); // 创建掩膜,目的是为了隔离nan值的干扰 mask = cv::Mat::zeros(phase2.size(), CV_8UC1); mask.setTo(255, phase2 == phase2); // 初始化三通道颜色图 cv::Mat color1, color2, color3; color1 = cv::Mat::zeros(temp.size(), temp.type()); color2 = cv::Mat::zeros(temp.size(), temp.type()); color3 = cv::Mat::zeros(temp.size(), temp.type()); // 基于灰度图的灰度层级,给其上色,最底的灰度值0为蓝色(255,0,0),最高的灰度值255为红色(0,0,255),中间的灰度值127为绿色(0,255,0) for (int i = 0; i < row; ++i) { uchar *c1 = color1.ptr<uchar>(i); uchar *c2 = color2.ptr<uchar>(i); uchar *c3 = color3.ptr<uchar>(i); uchar *r = temp.ptr<uchar>(i); uchar *m = mask.ptr<uchar>(i); for (int j = 0; j < col; ++j) { if (m[j] == 255) { if (r[j] > (3 * 255 / 4) && r[j] <= 255) { c1[j] = 255; c2[j] = 4 * (255 - r[j]); c3[j] = 0; } else if (r[j] <= (3 * 255 / 4) && r[j] > (255 / 2)) { c1[j] = 255 - 4 * (3 * 255 / 4 - r[j]); c2[j] = 255; c3[j] = 0; } else if (r[j] <= (255 / 2) && r[j] > (255 / 4)) { c1[j] = 0; c2[j] = 255; c3[j] = 4 * (255 / 2 - r[j]); } else if (r[j] <= (255 / 4) && r[j] >= 0) { c1[j] = 0; c2[j] = 255 - 4 * (255 / 4 - r[j]); c3[j] = 255; } else { c1[j] = 0; c2[j] = 0; c3[j] = 0; } } } } // 三通道合并,得到颜色图 vector<cv::Mat> images; images.push_back(color3); images.push_back(color2); images.push_back(color1); cv::merge(images, result); return result; }
测试效果
图1 参考灰度图上色效果
图2 基于参考灰度图色板的上色效果
如上图所示,为了方便,我生成了一个2001*2001的图像矩阵,并设置了另一个对比图像,该图像为原图像的1/2,那么原图像就是参考灰度图,而对比图像就是需要基于参考灰度图色板上色的灰度图。图1为参考灰度图的上色效果,图2是基于参考灰度图色板给对比图像上色的效果图。原图像的数据从0-1.3左右,其颜色变化从蓝到绿再到红,而对比图像的数据从0-1.3/2左右,则颜色变化为蓝到绿,满足了前面提到的需求。