使用 cv::fisheye::undistortImage 去扭曲鱼眼图像

Question

我正在尝试使用 OpenCV 使鱼眼图像不失真。我从相机的内存中获得了相机矩阵和畸变系数。我假设它们是准确的。正如您在下面的代码中看到的，我使用的是cv::fisheye::undistortImage。我发现这个 GitHub Issue post 声称它应该可以工作，但是，未失真的图像框架看起来不正确！所以，很明显有些东西不起作用。

代码如下：

#include <opencv2/highgui.hpp>
#include <opencv2/calib3d.hpp>
#include <opencv2/core/mat.hpp>

int main()
{
    cv::Mat cameraMatrix = cv::Mat(3,3, CV_64F, double(0));
    cv::Mat distortionCoeffs = cv::Mat(1,4, CV_64F, double(0));

    cameraMatrix.at<double>(0, 0) = 286.7037963867188;
    cameraMatrix.at<double>(0, 1) = 0;
    cameraMatrix.at<double>(0, 2) = 413.3463134765625;
    cameraMatrix.at<double>(1, 0) = 0;
    cameraMatrix.at<double>(1, 1) = 286.7817993164062;
    cameraMatrix.at<double>(1, 2) = 397.1785888671875;
    cameraMatrix.at<double>(2, 0) = 0;
    cameraMatrix.at<double>(2, 1) = 0;
    cameraMatrix.at<double>(2, 2) = 1;

    distortionCoeffs.at<double>(0,0) = -0.01078350003808737;
    distortionCoeffs.at<double>(0,1) = 0.04842806980013847;
    distortionCoeffs.at<double>(0,2) = -0.04542399942874908;
    distortionCoeffs.at<double>(0,3) = 0.008737384341657162;

    cv::Mat input_frame = cv::imread("fisheye_input.png");
    cv::Mat output_frame;

    cv::fisheye::undistortImage(input_frame,output_frame,cameraMatrix,distortionCoeffs, cv::noArray(), cv::Size(input_frame.cols,input_frame.rows));    

    cv::imshow("Input Image", input_frame);
    cv::imshow("Output Image", output_frame);
    cv::waitKey(-1);
    return 0;
}

代码输出：

如果您想自己尝试一下，这里是原始鱼眼图像：

Answer 1

我无法准确找到您的代码中的真正问题。当我阅读documentation 时，我看到了：

该功能只是一个组合 fisheye::initUndistortRectifyMap （使用 unity R ）和 remap （使用 双线性插值）。

当我深入研究使用这两种功能组合（initUndistortRectifyMap 和remap）而不是使用undistortImage 时，我发现@Micka 回答了this post。我试过了，效果很好，下面是代码和结果：

#include <opencv2/highgui.hpp>
#include <opencv2/calib3d.hpp>
#include <opencv2/core/mat.hpp>

#include <opencv2/imgproc/imgproc.hpp>
int main()
{
    cv::Mat cameraMatrix = cv::Mat(3,3, cv::DataType<double>::type);
    cv::Mat distortionCoeffs = cv::Mat(4,1, cv::DataType<double>::type);

    cameraMatrix.at<double>(0, 0) = 286.7037963867188;
    cameraMatrix.at<double>(0, 1) = 0;
    cameraMatrix.at<double>(0, 2) = 413.3463134765625;
    cameraMatrix.at<double>(1, 0) = 0;
    cameraMatrix.at<double>(1, 1) = 286.7817993164062;
    cameraMatrix.at<double>(1, 2) = 397.1785888671875;
    cameraMatrix.at<double>(2, 0) = 0;
    cameraMatrix.at<double>(2, 1) = 0;
    cameraMatrix.at<double>(2, 2) = 1;

    distortionCoeffs.at<double>(0,0) = -0.01078350003808737;
    distortionCoeffs.at<double>(1,0) = 0.04842806980013847;
    distortionCoeffs.at<double>(2,0) = -0.04542399942874908;
    distortionCoeffs.at<double>(3,0) = 0.008737384341657162;

    cv::Mat E = cv::Mat::eye(3, 3, cv::DataType<double>::type);

    cv::Mat input_frame = cv::imread("fishEye.png");

    cv::Size size = { input_frame.cols, input_frame.rows };

    cv::Mat map1;
    cv::Mat map2;
    //cv::fisheye::undistortImage(input_frame,output_frame,cameraMatrix,distortionCoeffs, E, cv::Size(input_frame.cols,input_frame.rows));

    cv::fisheye::initUndistortRectifyMap(cameraMatrix, distortionCoeffs, E, cameraMatrix, size, CV_16SC2, map1, map2);

    cv::Mat undistort;

    cv::remap(input_frame, undistort, map1, map2, cv::INTER_LINEAR,
              CV_HAL_BORDER_CONSTANT);

    cv::imshow("Input Image", input_frame);
    cv::imshow("Output Image", undistort    );
    cv::waitKey(-1);
    return 0;
}

输出：