BTW
The above animated gif should serve as a demo video to identify objects and close and distant layers
Hi,
I am looking for step-by-step manual on live obstacle avoidance algorithm based on depth maps live generated from 2 cameras.
There is a number of depth map generator apps, taking as input 2 still images.
I would like to install 2 webcams to PC ( MS windows) and have 2 individual left eye and right eye frames captured,
processed by depth map generator, generate depth layers (close, far), use close layer in depth map to calculate obstacle's topology and take action and to loop the above steps.
Another option is to use Kinect v1 or v2 ( complicated installation of drivers for MS Windows)
or latest smartphone coming with twin cameras/ lens + depth map code.
As a demo I am looking for depth map application run in HTML5, getting live video stream from 2 cameras as input.
HTML5 is slow and heavy but code can be scripted.
Problem with depth maps generated from low-res images are artifacts generated, affecting correct close/ far obstacle identification (Kinect and others).
Please let me know your opinion
Replies