为什么我们需要将（camera_matrix 1& 2）和（R和T）参数输入到stereoRectify（）？ [英] Why do we need to input both (camera_matrix 1 & 2) and (R and T) params to stereoRectify()?

问题描述

要进行立体声校准，我们需要执行3个步骤（输入参数 - >输出参数）：

1. calibrateCamera（） ：obj_corners，img_corners - > camera_matrix，distortion_coeffs




2. stereoCalibrate（） ：obj_corners，img_corners1 / 2，camera_matrix1 / 2，distortion_coeffs1 / 2 - > R，T，E，F




3. stereoRectify（） ：camera_matrix1 / 2，distortion_coeffs1 / 2，R，T - > R1 / 2，P1 / 2，Q

我错过了一些参数。

现在我们可以使用Q for reprojectImageTo3D（） 和 triangulatePoints（） 。

但是为什么我们需要输入（camera_matrix 1& 2）和（R& T）参数转换为 stereoRectify（） ？

已知 camera_matrix 已经包含R& T：

  struct CV_EXPORTS CameraParams 
 {
 CameraParams 
 CameraParams（const CameraParams& other）; 
 const CameraParams& operator =（const CameraParams& other）; 
 Mat K（）const; 
 
 double focal; //焦距
双方面; // Aspect ratio 
 double ppx; // Principal X 
 double ppy; //主点Y 
 Mat R; // Rotation 
 Mat t; // Translation 
}; 
  

T，我们从 calibrateCamera（） 和R& T，它包含在CameraParams中，我们从 c>

<％> 20F，％20TermCriteria％20criteria，％20int％20flags） =nofollow div class =h2_lin>解决方案

这里的问题是不一致的术语。 OpenCV调用相机矩阵是包含内在函数的3x3矩阵 K 。 Harley和Zisserman调用相机矩阵是3x4相机投影矩阵 P = K * [R | t] ，其包括内在函数和extrinsics。

To do stereo calibrate we need to do 3 steps (input params --> output params):

1. calibrateCamera(): obj_corners, img_corners --> camera_matrix, distortion_coeffs

2. stereoCalibrate(): obj_corners, img_corners1/2, camera_matrix1/2, distortion_coeffs1/2 --> R, T, E, F

3. stereoRectify(): camera_matrix1/2, distortion_coeffs1/2, R, T --> R1/2, P1/2, Q

Some params I missed.

And now we can use Q for reprojectImageTo3D() and P1 or P2 for triangulatePoints().

But why do we need to input both (camera_matrix 1 & 2) and (R & T) params to stereoRectify()?

As known camera_matrix already contain R & T:

struct CV_EXPORTS CameraParams
{
    CameraParams();
    CameraParams(const CameraParams& other);
    const CameraParams& operator =(const CameraParams& other);
    Mat K() const;

    double focal; // Focal length
    double aspect; // Aspect ratio
    double ppx; // Principal point X
    double ppy; // Principal point Y
    Mat R; // Rotation
    Mat t; // Translation
};

What's the difference between R & T which we get from calibrateCamera() and R & T that contained in CameraParams which we get from stereoCalibrate()?

解决方案

The problem here is inconsistent terminology. What OpenCV calls camera matrix is a 3x3 matrix K containing the intrinsics. What Harley and Zisserman call camera matrix is a 3x4 camera projection matrix P = K * [R|t], which includes both the intrinsics and the extrinsics.

这篇关于为什么我们需要将（camera_matrix 1&amp; 2）和（R和T）参数输入到stereoRectify（）？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！