Compute First features, then matching

[danielsanjosepro]

Goal

~Close 40, 39, 38~

• [x] Compute features first of all the images
• [ ] ~Create dense reconstruction with epipolar rectification~
• [ ] ~Recover from a wrongly computed E matrix~

[danielsanjosepro]

Delivery Area, two images, 27051 points. Some work on progress

[danielsanjosepro]

Harder than I though. There is some kind of distortion with the images. My laptop cant handle these point clouds anyway ( 47398 points )

[lufixSch]

How did you implement that this fast?

[lufixSch]

Harder than I though. There is some kind of distortion with the images. My laptop cant handle these point clouds anyway ( 47398 points )

Looks promising nevertheless. We could take a look at it with my Workstation. Maybe it can handle such a high number of points. We still might have to implement some form of Filter to decide which features/pixels are interesting before reconstructing every single pixel.