Brief introduction of realsense d435i imaging principle

Writing graduation thesis involves realsense d435i The imaging principle of the camera , After searching and comparing , The following conclusions are drawn .

First, let's take a look at the structure of this camera ：

The picture is from ：https://blog.csdn.net/lemonxiaoxiao/article/details/109297834

1、 Principle of color image imaging

RGB Monocular pinhole model of camera , The image below ：

2、 Depth image imaging principle

Images based on left and right infrared cameras , use Triangulation Calculation . Because there are few feature points in some scenes , Therefore, the infrared dot matrix projector emits infrared dot matrix patterns into the scene （ People can't see , Infrared camera can capture ）, Add feature points , Improve the imaging effect .
ROS The depth image published in is the same as that in the figure Right infrared camera The alignment of the , Left infrared 、 Right infrared 、 The depth map is shown in the following figure ： 

in addition ,realsense-ROS Released left infrared 、 Right infrared 、 Depth map 、 Internal parameters of color camera , And the internal parameters of the depth map aligned with the color map ：

1、 Right infrared camera parameters 
D: [0.0, 0.0, 0.0, 0.0, 0.0]
K: [378.98150634765625, 0.0, 318.2408447265625, 0.0, 378.98150634765625, 241.04586791992188, 0.0, 0.0, 1.0]
R: [1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0]
P: [378.98150634765625, 0.0, 318.2408447265625, 0.0, 0.0, 378.98150634765625, 241.04586791992188, 0.0, 0.0, 0.0, 1.0, 0.0]

2、 Left infrared camera parameters 
D: [0.0, 0.0, 0.0, 0.0, 0.0]
K: [378.98150634765625, 0.0, 318.2408447265625, 0.0, 378.98150634765625, 241.04586791992188, 0.0, 0.0, 1.0]
R: [1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0]
P: [378.98150634765625, 0.0, 318.2408447265625, 18.91739845275879, 0.0, 378.98150634765625, 241.04586791992188, 0.0, 0.0, 0.0, 1.0, 0.0]

3、 Depth image parameters 
D: [0.0, 0.0, 0.0, 0.0, 0.0]
K: [378.98150634765625, 0.0, 318.2408447265625, 0.0, 378.98150634765625, 241.04586791992188, 0.0, 0.0, 1.0]
R: [1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0]
P: [378.98150634765625, 0.0, 318.2408447265625, 0.0, 0.0, 378.98150634765625, 241.04586791992188, 0.0, 0.0, 0.0, 1.0, 0.0]

4、 Color camera parameters 
D: [0.0, 0.0, 0.0, 0.0, 0.0]
K: [613.4265747070312, 0.0, 329.3377990722656, 0.0, 613.5404663085938, 243.68722534179688, 0.0, 0.0, 1.0]
R: [1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0]
P: [613.4265747070312, 0.0, 329.3377990722656, 0.0, 0.0, 613.5404663085938, 243.68722534179688, 0.0, 0.0, 0.0, 1.0, 0.0]

5、 Parameters of depth aligned color map 
D: [0.0, 0.0, 0.0, 0.0, 0.0]
K: [613.4265747070312, 0.0, 329.3377990722656, 0.0, 613.5404663085938, 243.68722534179688, 0.0, 0.0, 1.0]
R: [1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0]
P: [613.4265747070312, 0.0, 329.3377990722656, 0.0, 0.0, 613.5404663085938, 243.68722534179688, 0.0, 0.0, 0.0, 1.0, 0.0]

It can be seen that , The parameters of the depth image are the same as those of the right infrared camera , It can also be proved that the depth image is consistent with Right infrared camera alignment . in addition , Parameters of depth map aligned with color map The parameters are the same as those of the color camera .

It is worth noting that ,ROS Depth images are also released to color cameras 、 Left infrared 、 External parameter of right infrared ：

6、depth->color  External reference 
rotation: [0.9999861717224121, 0.0002028869348578155, -0.005260064732283354, -0.0001973225735127926, 0.9999994039535522, 0.0010583496186882257, 0.005260276608169079, -0.0010572969913482666, 0.9999856352806091]
translation: [0.014868799597024918, -3.15127799694892e-05, 0.0001727673807181418]

7、depth-> Right infrared   External reference 
rotation: [1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0]
translation: [0.0, 0.0, 0.0]

8、depth-> Left infrared   External reference 
rotation: [1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0]
translation: [0.0, 0.0, 0.0]

There is no problem with the external parameters from the depth map to the color camera , But the external parameters of the two infrared cameras are the same , According to the above conclusion , The external parameter from the depth map to the right infrared is the unit rotation matrix +0 translation , But the external parameter to the left infrared should not 0 translation （ I didn't think clearly why ）.