Eight of image processing technology: cloud point, the depth image

The depth image

• Also called from the image means from the image acquirer scene points to the distance (depth) value as the pixel value .

• Acquisition method are: LIDAR depth imaging, computed stereoscopic binocular vision imaging , the coordinate measuring machine method, a moire method, a structured light method .

Point cloud

• When a beam of laser irradiation when the surface of an object, the reflected laser light will carry with the azimuth , distance information. If the laser beam according to a certain trajectory for scanning laser light reflected by the point information, it will be recorded while scanning, since the scanning is extremely fine , it is possible to obtain a large number of laser spots , thus can form a laser point cloud . Cloud point format * .las ; *. PCD ; * .txt like.

The depth image through coordinate conversion can be calculated as the point cloud data ; rules and information necessary to point cloud data can be back-calculated to the depth image

 

RGB-D image provides rgb image pixel coordinates in the x, y coordinates, and provides a depth map Z ZZ directly coordinates of the camera coordinate system, i.e. the distance from the camera point.

The RGB-D image information and camera internal reference, coordinates can be calculated at any pixel a camera coordinate system.

The parameters involved in the external RGB-D image information and camera can calculate the coordinates of any pixel in the world coordinate system.

The camera field of view, the coordinates of the obstacle point in the camera coordinate system is the point cloud data of the sensor, i.e. the point cloud data in the camera coordinate system. Sensor point cloud data can be calculated from the coordinates of the internal reference and the RGB-D camera image provided.

The coordinates of the point of obstruction in all the world coordinate system, the map is the point cloud data, that is, the point cloud data in the world coordinate system. Point cloud map data may be calculated based on the coordinates provided by the RGB-D image with the camera and the external reference internal reference.

https://blog.csdn.net/go_clever_boy/article/details/102295371?depth_1-utm_source=distribute.pc_relevant.none-task&utm_source=distribute.pc_relevant.none-task

https://blog.csdn.net/ff_xun/article/details/85318093

 

References
"robotics, machine vision and control"

https://www.cnblogs.com/gaoxiang12/p/4652478.html

http://www.cnblogs.com/cv-pr/p/5719350.html