基于opencv的一维Fourier变换

此程序主要实现了fft、ifft计算。原本是用于cpu雷达信号脉冲压缩的C++实现。程序调试后无错误，可以运行，程序需要读取磁盘中的.dat文件实现数据导入。

调试环境：VS2015+opencv

//2016.11.2 @Xidian//DSP雷达信号处理2//调用opencv中dft()函数实现fft和ifft#include<iostream>#include<fstream>#include<opencv2\opencv.hpp>using namespace std;using namespace cv;const int N = 7680;const int M = 8192;void main(){    double count1 = getTickCount();    //-------------------对回波信号echo进行8192点FFT-----------------       FILE* fp;    fp=fopen ("echo.dat", "r");    float temp[7680 * 2] = {0};    vector<float> echo;  //    for (int i = 0; i < 7680*2; i++)    {        fscanf(fp, "%f", &temp[i]);        echo.push_back(temp[i]);    }    fclose(fp);    //echo信号的实部、虚部    Mat_<float> echo_real = Mat::zeros(1, 8192, CV_32F);    Mat_<float> echo_img = Mat::zeros(1, 8192, CV_32F);    vector<float>hh[2];    for (int i = 0; i < 7680 * 2; i += 2)    {        hh[0].push_back(echo[i]);        hh[1].push_back(echo[i + 1]);    }    for (int i = 0; i < 7680; i++)    {        echo_real.at<float>(i) = hh[0][i];        echo_img.at<float>(i) = hh[1][i];    }    //将echo的实、虚部组合为复数complex    vector<Mat_<float>> data;    data.push_back(echo_real);    data.push_back(echo_img);    Mat complex;    merge(data, complex);    //计算FFT    dft(complex, complex);    //将echo信号的8192点fft写入.dat文件    vector<Mat_<float>> echo_fft;    split(complex, echo_fft);    FILE* fp1;    FILE* fp2;    fp1 = fopen("echo_fft_real.dat", "w");    fp2 = fopen("echo_fft_img.dat", "w");    for (int i = 0; i < 8192; i++)    {        //将echo进行FFT后的实部写入文本        fprintf(fp1, "%f\n", echo_fft[0].at<float>(i));        //将echo进行FFT后的虚部写入文本        fprintf(fp2, "%f\n", echo_fft[1].at<float>(i));    }    //-----------------对脉压系数coeff进行8192点FFT-----------------    //Mat_<float> coeff_real = Mat::zeros(1, 8192,CV_32F);  //coeff的实部    //Mat_<float> coeff_img = Mat::zeros(1, 8192, CV_32F);  //coeff的虚部    //Mat_<float> coeff_complex;                              //coeff的复数    //Mat_<float> coeff_fft = Mat::zeros(1, 8192, CV_32F);  //coeff的8192点FFT    //    //FILE* fp3;    //fp3 = fopen("coeff.dat", "r");    //float temp3[7680 * 2];    //for (int i = 0; i <7680*2; i++)    //{ //读取coeff.dat文本数据    //  fscanf(fp3, "%f", &temp3[i]);    //}    //vector<float> GG[2];    //for (int i = 0; i < 7680*2; i+=2)    //{    //  GG[0].push_back(temp3[i]);    //  GG[1].push_back(temp3[i + 1]);    //}    //for (int i = 0; i < 7680; i++)    //{    //  coeff_real.at<float>(i) = GG[0][i];    //  coeff_img.at<float>(i) = GG[1][i];    //}    //vector<Mat_<float>>coeff;    //coeff.push_back(coeff_real);    //coeff.push_back(coeff_img);    ////组合复数    //merge(coeff, coeff_complex);    ////FFT计算    //dft(coeff_complex, coeff_complex);    //    //----------------------读取coeff_fft并保存为复数---------------------------    Mat_<float> coeff_fftReal2 = Mat::zeros(1, 8192,CV_32F);  //coeff_fft的实部    Mat_<float> coeff_fftImg2 = Mat::zeros(1, 8192, CV_32F);  //coeff_fft的虚部    Mat coeff_fftComplex2;                            //coeff_fft的复数    FILE* fp4;    fp4 = fopen("coeff_fft.dat", "r");    float temp4[7680 * 2] = {0};    for (int i = 0; i < 7680*2; i++)    {        fscanf(fp4, "%f", &temp4[i]);    }    vector<float> JJ[2];    for (int i = 0; i < 7680*2; i+=2)    {        JJ[0].push_back(temp4[i]);        JJ[1].push_back(temp4[i+1]);    }    for (int i = 0; i < 7680; i++)    {        coeff_fftReal2.at<float>(i) = JJ[0][i];        coeff_fftImg2.at<float>(i) = JJ[1][i];    }    vector<Mat_<float>>coeff_fft;    coeff_fft.push_back(coeff_fftReal2);    coeff_fft.push_back(coeff_fftImg2);    //组合为复数    merge(coeff_fft, coeff_fftComplex2);    //------------------频域相乘-----------------------------    Mat_<float> pc_freq0Img;    Mat_<float> pc_freq0Real;    Mat pc_freq0Complex;    vector<Mat_<float>> pc_freq0;    for (int i = 0; i < 8192; i++)    {        Vec2f a = complex.at<Vec2f>(i);        Vec2f b = coeff_fftComplex2.at<Vec2f>(i);        float real = a[0] * b[0] - a[1] * b[1];           float img = a[0] * b[1] + a[1] * b[0];        img = -1.0*img;    /*  cout << "real=" << real << endl;        cout << "imag=" << img << endl;*/        pc_freq0Real.push_back(real);        pc_freq0Img.push_back(img);   //共轭    }    //组合为复数    pc_freq0.push_back(pc_freq0Real);    pc_freq0.push_back(pc_freq0Img);    merge(pc_freq0, pc_freq0Complex);    /*cout << pc_freq0Complex << endl;*/    //---------------------IFFT计算-----------------------    //直接计算ifft    dft(pc_freq0Complex, pc_freq0Complex, DFT_SCALE);  //???与matlab计算结果相差太大    /*cout << pc_freq0Complex << endl;*/    //用fft计算ifft    //step1 计算fft    //dft(pc_freq0Complex, pc_freq0Complex);    ////step2 共轭、乘以1/N    //for (int i = 0; i < 8192; i++)    //{    //  Vec2f bb = pc_freq0Complex.at<Vec2f>(i);    //  float real2 = bb[0];    //  float imag2 = bb[1];    //  imag2 = imag2*(-1.0);    //  pc_freq0Complex.at<Vec2f>(i) = (1.0/8192.0)*Vec2f(real2, imag2);    //}    //    //cout << pc_freq0Complex << endl;    //------------------计算结果写入.dat文件(包含暂态点)------------------------    FILE* fp5;    FILE* fp6;    fp5 = fopen("pc_freq1_real.dat","w");    fp6 = fopen("pc_freq1_imag.dat","w");    for (int i = 0; i < 8192; i++)    {        Vec2f c = pc_freq0Complex.at<Vec2f>(i);        /*cout << "pc_freq0.real=" << c[0]<<endl;        cout << "pc_freq0.imag=" << c[1]<<endl;*/        fprintf(fp5, "%f\n",c[0]);        fprintf(fp6, "%f\n", c[1]);    }    fclose(fp5);    fclose(fp6);    double count2 = getTickCount();    double freq = getTickFrequency();    double time = (count2 - count1) / freq;    cout << "运行时间=" << time << endl;    cout << "Successful!" << endl;    system("PAUSE");}