OpenCvSharp4.6 编写的 C# 代码版本 // 求取极值点 void Caliper::FindExtremePoint() { // 1. 检查搜索路径像素数据是否为空 if(pathPixelValueAfterFilter.empty()) return; // 2. 遍历滤波后的数据，查找极值点 extremePoints.clear(); switch (polarity) { // 不考虑极性 case 0 : for(size_t i = 1; i < pathPixelValueAfterFilter.size() - 1; ++i) { // 两边异号为极值点 if (size_t(std::abs(pathPixelValueAfterFilter.at(i))) >= contrastThreshold && (0 > (pathPixelValueAfterFilter.at(i) - pathPixelValueAfterFilter.at(i - 1)) * (pathPixelValueAfterFilter.at(i + 1) - pathPixelValueAfterFilter.at(i))) ) { extremePoints.push_back(ExtremPointInfo(path.at(i), static_cast<int>(i), static_cast<int>(pathPixelValueAfterFilter.at(i)))); } } break; // 黑到白 case 1 : for(size_t i = 1; i < pathPixelValueAfterFilter.size() - 1; ++i) { // 两边异号为极值点 if (pathPixelValueAfterFilter.at(i) > 0 && size_t(std::abs(pathPixelValueAfterFilter.at(i))) >= contrastThreshold && (0 > (pathPixelValueAfterFilter.at(i) - pathPixelValueAfterFilter.at(i - 1)) * (pathPixelValueAfterFilter.at(i + 1) - pathPixelValueAfterFilter.at(i))) ) { extremePoints.push_back(ExtremPointInfo(path.at(i), static_cast<int>(i), static_cast<int>(pathPixelValueAfterFilter.at(i)))); } } break; // 白到黑 case -1: for(size_t i = 1; i < pathPixelValueAfterFilter.size() - 1; ++i) { // 两边异号为极值点 if (pathPixelValueAfterFilter.at(i) < 0 && size_t(std::abs(pathPixelValueAfterFilter.at(i))) >= contrastThreshold && (0 > (pathPixelValueAfterFilter.at(i) - pathPixelValueAfterFilter.at(i - 1)) * (pathPixelValueAfterFilter.at(i + 1) - pathPixelValueAfterFilter.at(i))) ) { extremePoints.push_back(ExtremPointInfo(path.at(i), static_cast<int>(i), static_cast<int>(pathPixelValueAfterFilter.at(i)))); } } break; default: assert(false); break; }OpenCvSharp4.6 编写的 C# 代码版本函数

这段代码是用于在图像中搜索极值点的，主要包括以下步骤：

1. 检查搜索路径像素数据是否为空；

2. 遍历滤波后的数据，查找极值点；

3. 根据极性类型分别处理不同的情况，包括不考虑极性、黑到白和白到黑；

4. 找到极值点后将其信息保存到 `extremePoints` 中。

其中，`pathPixelValueAfterFilter` 是经过滤波后的搜索路径像素数据，`contrastThreshold` 是对比度的阈值，`path` 是搜索路径的像素坐标，`polarity` 是极性类型。

相关问题

// 极值点评分 void Caliper::ExtremePointRating() { if(extremePoints.empty()) return; // 评分： value * (1 - |i - 0.5 * le| / 0.5 * len) int halfLen = static_cast<int>(path.size() / 2); for(auto& e: extremePoints) { e.score = e.value * (1 - static_cast<double>(std::abs(e.ps - halfLen)) / halfLen); } //按评分排序：降序 sort(extremePoints.begin(), extremePoints.end(), CompareExtremPointInfoBysocre); result.peak = extremePoints.at(0).pt; if(extremePoints.size() == 1) { result.center = result.front = result.back = extremePoints.at(0).pt; return; } if(extremePoints.size() == 2) { int a = std::abs(extremePoints.at(0).ps - halfLen); int b = std::abs(extremePoints.at(1).ps - halfLen); if(a > b) { result.front = result.back = extremePoints.at(0).pt; result.center = extremePoints.at(1).pt; }else { } }else { std::vector<ExtremPointInfo> epf(extremePoints.begin(), extremePoints.begin() + 3); sort(epf.begin(), epf.end(), CompareExtremPointInfoByPs);//降序 result.back = epf.at(0).pt; result.center = epf.at(1).pt; result.front = epf.at(2).pt; } }//ExtremePointRating 基于opencvsharp4.6 请优化代码,生成新的函数

为了优化代码，可以考虑以下几个方面：

1. 减少不必要的类型转换

在原有代码中，存在如下的类型转换：

int halfLen = static_cast<int>(path.size() / 2);

可以改为使用整数除法运算符 `/`，避免类型转换的开销：

int halfLen = path.size() / 2;

2. 避免不必要的容器复制

在原有代码中，存在如下的容器复制操作：

std::vector<ExtremPointInfo> epf(extremePoints.begin(), extremePoints.begin() + 3);

可以改为直接使用 `std::array`，避免不必要的容器复制：

std::array<ExtremPointInfo, 3> epf = {extremePoints[0], extremePoints[1], extremePoints[2]};

3. 使用 lambda 表达式简化代码

在原有代码中，使用了函数指针作为 `sort` 函数的比较函数：

sort(extremePoints.begin(), extremePoints.end(), CompareExtremPointInfoBysocre);
sort(epf.begin(), epf.end(), CompareExtremPointInfoByPs);

可以改为使用 lambda 表达式，使代码更加简洁：

sort(extremePoints.begin(), extremePoints.end(), [](const ExtremPointInfo& a, const ExtremPointInfo& b) { return a.score > b.score; });
sort(epf.begin(), epf.end(), [](const ExtremPointInfo& a, const ExtremPointInfo& b) { return a.ps > b.ps; });

基于以上优化，可以得到如下的新函数：

void Caliper::ExtremePointRatingOptimized()
{
    if (extremePoints.empty())
        return;

    // 评分： value * (1 - |i - 0.5 * le| / 0.5 * len)
    int halfLen = path.size() / 2;
    for (auto& e : extremePoints)
    {
        e.score = e.value * (1 - static_cast<double>(std::abs(e.ps - halfLen)) / halfLen);
    }

    // 按评分排序：降序
    sort(extremePoints.begin(), extremePoints.end(), [](const ExtremPointInfo& a, const ExtremPointInfo& b) { return a.score > b.score; });

    result.peak = extremePoints[0].pt;

    if (extremePoints.size() == 1)
    {
        result.center = result.front = result.back = extremePoints[0].pt;
        return;
    }

    if (extremePoints.size() == 2)
    {
        int a = std::abs(extremePoints[0].ps - halfLen);
        int b = std::abs(extremePoints[1].ps - halfLen);
        if (a > b)
        {
            result.front = result.back = extremePoints[0].pt;
            result.center = extremePoints[1].pt;
        }
        else
        {
            result.front = result.back = extremePoints[1].pt;
            result.center = extremePoints[0].pt;
        }
    }
    else
    {
        std::array<ExtremPointInfo, 3> epf = {extremePoints[0], extremePoints[1], extremePoints[2]};
        // 降序
        sort(epf.begin(), epf.end(), [](const ExtremPointInfo& a, const ExtremPointInfo& b) { return a.ps > b.ps; });

        result.back = epf[0].pt;
        result.center = epf[1].pt;
        result.front = epf[2].pt;
    }
}

bool searchDir_left2right //从左往右找点 public void FindExtremePoint(Mat image, bool searchDir_left2right , ref List<Point2f> zeroPList) { if (image.Width == 0) { return;// throw new ArgumentNullException(nameof(image)); } // 预处理：中值滤波 Cv2.MedianBlur(image, image, 3); // 计算极值和位置 Point2f extremaLoc; double extremaVal; double minVal, maxVal; var criteria = new TermCriteria(CriteriaTypes.Eps | CriteriaTypes.Count, 20, 0.03); //MinMaxLoc(InputArray src, out double minVal, out double maxVal, out Point minLoc, out Point maxLoc, InputArray? mask = null); Cv2.MinMaxLoc(image, out minVal, out maxVal, out Point minLoc, out Point _); Point2f[] corners = new[] { new Point2f(minLoc.X, minLoc.Y) }; Cv2.CornerSubPix(image, corners, new Size(3, 3), new Size(-1, -1), criteria); extremaVal = minVal; extremaLoc = corners[0]; zeroPList.Add( extremaLoc); } 基于opencvsharp4.6编写c#代码,生成新的函数

以下是基于OpenCvSharp4.6编写的C#函数：

public void FindExtremePoint(Mat image, bool searchDir_left2right, ref List<Point2f> zeroPList)
{
    if (image.Width == 0)
    {
        return;// throw new ArgumentNullException(nameof(image));
    }

    // 预处理：中值滤波
    Cv2.MedianBlur(image, image, 3);

    // 计算极值和位置
    Point2f extremaLoc;
    double extremaVal;
    double minVal, maxVal;
    var criteria = new TermCriteria(CriteriaTypes.Eps | CriteriaTypes.Count, 20, 0.03);

    //MinMaxLoc(InputArray src, out double minVal, out double maxVal, out Point minLoc, out Point maxLoc, InputArray? mask = null);
    Cv2.MinMaxLoc(image, out minVal, out maxVal, out Point minLoc, out Point _);

    Point2f[] corners = new[] { new Point2f(minLoc.X, minLoc.Y) };
    Cv2.CornerSubPix(image, corners, new Size(3, 3), new Size(-1, -1), criteria);

    extremaVal = minVal;
    extremaLoc = corners[0];

    zeroPList.Add(extremaLoc);
}

此函数的作用是从输入的图像中查找极值点，并将其位置信息存储在一个列表中。函数的参数包括：

- image：输入的图像
- searchDir_left2right：一个bool值，表示是否从左往右查找点
- zeroPList：一个Point2f类型的列表，用于存储极值点的位置信息。

函数内部首先进行了中值滤波的预处理，然后使用MinMaxLoc函数查找图像中的最小值和最大值，并将最小值的位置信息存储在corners数组中。接着使用CornerSubPix函数对corners数组进行亚像素级别的精细化处理，得到最终的极值点位置信息extremaLoc。最后将extremaLoc加入到zeroPList列表中。