For ORB pyramid, anyone have some references/explanation on the code segment at https://github.com/Itseez/opencv/blob...
I have downloaded books, lecture notes, papers, and articles to read, but I could not understand the maths involved in ORB pyramid. Any ideas ?
Thanks !
The pyramid makes a set of reduced sized images (a pyramid) and performs ORB on each of them. That lets you track the points across scale changes, where the object gets closer or farther from the camera. For example, you find a point on the first, unresized image. Then the object gets closer and you will find the same keypoint on the second layer of the pyramid because it's larger.
Generally, blur and resize down. The factor by which you resize is variable, with 0.8 and 0.5 being popular.
Asked: 2016-05-15 08:53:40 -0600
Seen: 1,024 times
Last updated: May 15 '16
