3D reconstruction system | L2 camera model

One 、 Pinhole camera model

1. External parameter matrix

Imaging phase I ： Coordinate transformation from world coordinate system to camera coordinate system

• Change the way ：
  among R Express rotate matrix ,t Express translation matrix
  

2. Internal parameter matrix

Imaging phase II ： The camera coordinate system is converted to the normalized image plane coordinate system

Be careful ! The coordinates obtained here are in millimetres

Imaging phase III ：、 The normalized image plane coordinate system is converted to the physical image plane coordinate system

Here we will present the effect on the final plane to the pixel plane

in addition , In the physical image plane coordinate system , Origin in the upper left corner , In this case, coordinate system translation is required

Finally, the coordinates of the object under the image plane are obtained from the world coordinates

3. Perspective matrix

Integrate the above phases , The perspective matrix is obtained from the inner matrix and the outer matrix

4. Radial distortion

1. Cause of distortion

Due to the imaging rules of the large and the small , The ideal pinhole camera model can not get completely orthogonal imaging results under wide-angle lens , Therefore, it is necessary to correct the distortion of the image plane results .

2. Distortion correction

First, we need to estimate the distortion coefficient （k1,k2…kn）

Then solve the equation to get the correct coordinates of the object under distortion