常用算法小记

C++与Qt实现：标准差滤波与线性拟合

最新推荐文章于 2025-01-08 17:36:12 发布

hebbely

最新推荐文章于 2025-01-08 17:36:12 发布

阅读量633

点赞数 1

CC 4.0 BY-SA版权

文章标签：算法 c++ 线性拟合标准方差滤波

本文链接：https://blog.csdn.net/hebbely/article/details/129691023

标准差阈值滤波

这是一个使用标准差阈值滤波器的C++实现。对于每个数据点，我们将计算其邻近点的标准差。如果该点的值与均值之差大于标准差的阈值倍数，则将其替换为其邻近点的平均值。

#include <iostream>
#include <vector>
#include <cmath>

std::vector<double> standard_deviation_filter(const std::vector<double>& data, int window_size) {
    std::vector<double> filtered_data;
    int n = data.size();
    filtered_data.reserve(n);

    for (int i = 0; i < n; i++) {
        int start = std::max(0, i - window_size / 2);
        int end = std::min(n - 1, i + window_size / 2);

        double sum = 0;
        double sum_square = 0;
        int count = end - start + 1;

        for (int j = start; j <= end; j++) {
            sum += data[j];
            sum_square += data[j] * data[j];
        }

        double mean = sum / count;
        double variance = (sum_square / count) - (mean * mean);
        double std_dev = std::sqrt(variance);
        double threshold = 2.0 * std_dev;

        if (std::abs(data[i] - mean) > threshold) {
            filtered_data.push_back(mean);
        } else {
            filtered_data.push_back(data[i]);
        }
    }

    return filtered_data;
}

int main() {
    std::vector<double> data = {33.741, 31.651, 30.298, 29.332, 28.582, 28.024, 27.584, 27.214,
                                26.924, 26.685, 26.471, 26.331, 26.166, 26.04,  25.948, 25.846,
                                25.762, 25.686, 25.618, 25.565, 25.535, 25.468, 25.43,  25.373,
                                25.339, 25.296, 25.26,  25.235, 25.21,  25.159, 25.103, 25.105,
                                25.071, 25.048, 25.015, 25.012, 24.985, 24.96,  24.902, 24.908,
                                24.894, 24.867, 24.817, 24.825, 24.806, 24.798, 24.745};

    int window_size = 3;
    std::vector<double> filtered_data = standard_deviation_filter(data, window_size);

    for (const auto& value : filtered_data) {
        std::cout << value << " ";
    }
    std::cout << std::endl;

    return 0;
}

在此示例中，我们将阈值设置为滑动窗口内标准差的2倍。请注意，动态阈值可能会导致过滤效果与固定阈值不同。您可以根据需要调整窗口大小和阈值倍数。

以下是使用Qt平台实现标准差滤波器的示例

#include <QCoreApplication>
#include <QVector>
#include <QDebug>
#include <cmath>

QVector<double> standard_deviation_filter(const QVector<double>& data, int window_size) {
    QVector<double> filtered_data;
    int n = data.size();
    filtered_data.reserve(n);

    for (int i = 0; i < n; i++) {
        int start = std::max(0, i - window_size / 2);
        int end = std::min(n - 1, i + window_size / 2);

        double sum = 0;
        double sum_square = 0;
        int count = end - start + 1;

        for (int j = start; j <= end; j++) {
            sum += data[j];
            sum_square += data[j] * data[j];
        }

        double mean = sum / count;
        double variance = (sum_square / count) - (mean * mean);
        double std_dev = std::sqrt(variance);
        double threshold = 2.0 * std_dev;

        if (std::abs(data[i] - mean) > threshold) {
            filtered_data.push_back(mean);
        } else {
            filtered_data.push_back(data[i]);
        }
    }

    return filtered_data;
}

int main(int argc, char *argv[])
{
    QCoreApplication a(argc, argv);

    QVector<double> data = {33.741, 31.651, 30.298, 29.332, 28.582, 28.024, 27.584, 27.214,
                            26.924, 26.685, 26.471, 26.331, 26.166, 26.04,  25.948, 25.846,
                            25.762, 25.686, 25.618, 25.565, 25.535, 25.468, 25.43,  25.373,
                            25.339, 25.296, 25.26,  25.235, 25.21,  25.159, 25.103, 25.105,
                            25.071, 25.048, 25.015, 25.012, 24.985, 24.96,  24.902, 24.908,
                            24.894, 24.867, 24.817, 24.825, 24.806, 24.798, 24.745};

    int window_size = 3;
    QVector<double> filtered_data = standard_deviation_filter(data, window_size);

    for (const auto& value : filtered_data) {
        qDebug() << value;
    }

    return a.exec();
}

移动平均滤波

以下是使用Qt平台实现平滑滤波器的示例

#include <QCoreApplication>
#include <QVector>
#include <QDebug>
#include <cmath>

QVector<double> moving_average_filter(const QVector<double>& data, int window_size) {
    QVector<double> filtered_data;
    int n = data.size();
    filtered_data.reserve(n);

    for (int i = 0; i < n; i++) {
        int start = std::max(0, i - window_size / 2);
        int end = std::min(n - 1, i + window_size / 2);

        double sum = 0;
        int count = end - start + 1;

        for (int j = start; j <= end; j++) {
            sum += data[j];
        }

        filtered_data.push_back(sum / count);
    }

    return filtered_data;
}

int main(int argc, char *argv[])
{
    QCoreApplication a(argc, argv);

    QVector<double> data = {33.741, 31.651, 30.298, 29.332, 28.582, 28.024, 27.584, 27.214,
                            26.924, 26.685, 26.471, 26.331, 26.166, 26.04,  25.948, 25.846,
                            25.762, 25.686, 25.618, 25.565, 25.535, 25.468, 25.43,  25.373,
                            25.339, 25.296, 25.26,  25.235, 25.21,  25.159, 25.103, 25.105,
                            25.071, 25.048, 25.015, 25.012, 24.985, 24.96,  24.902, 24.908,
                            24.894, 24.867, 24.817, 24.825, 24.806, 24.798, 24.745};

    int window_size = 3;
    QVector<double> filtered_data = moving_average_filter(data, window_size);

    for (const auto& value : filtered_data) {
        qDebug() << value;
    }

    return a.exec();
}

线性拟合

使用最小二乘法实现线性拟合

#include <iostream>
#include <vector>
#include <cmath>

using namespace std;

double linear_fit(vector<double> x, vector<double> y, double& a, double& b) {
    int n = x.size();
    double x_bar = 0.0, y_bar = 0.0, cov_xy = 0.0, var_x = 0.0;
    for (int i = 0; i < n; i++) {
        x_bar += x[i];
        y_bar += y[i];
    }
    x_bar /= n;
    y_bar /= n;
    for (int i = 0; i < n; i++) {
        cov_xy += (x[i] - x_bar) * (y[i] - y_bar);
        var_x += (x[i] - x_bar) * (x[i] - x_bar);
    }
    a = cov_xy / var_x;
    b = y_bar - a * x_bar;
    double residual_sum_of_squares = 0.0;
    for (int i = 0; i < n; i++) {
        double y_fit = a * x[i] + b;
        residual_sum_of_squares += (y[i] - y_fit) * (y[i] - y_fit);
    }
    double rmse = sqrt(residual_sum_of_squares / n);
    return rmse;
}

int main() {
    vector<double> x = {1.0, 2.0, 3.0, 4.0, 5.0};
    vector<double> y = {2.0, 4.0, 5.0, 4.5, 6.0};
    double a, b;
    double rmse = linear_fit(x, y, a, b);
    cout << "回归系数a: " << a << endl;
    cout << "截距b: " << b << endl;
    cout << "均方根误差RMSE: " << rmse << endl;
    return 0;
}