We are working on assessment of parameters to develop a more autonomous system that can adjust parameters according to the environment and task.

The following is a sample of the SVS interface displaying the left image of the stereo pair along with the disparity image produced.

Once the disparity is found between the two images, the 3D points can be generated. The following is a 3D view of the reconstructed points found using the SVS software.

Now I have 3D points, what is needed are the 3D surfaces. For this, we use Adam Hoover's OPUS code. We are considering changing to the Space Envelope Code, under strong suggestion by Adam.

The goal is to be able to use context to identify the type of environment being observed. For example, you could see what could be a table top or working surface of a desk. Behind that desk could be a chair. Usually, in that configuration, you would be an office environment. You identified the chair by only viewing the evidence of its back surface, oriented near the working surface.

To extend this further, you could identify a conference room configuration by viewing a table surface with multiple chairs (identified by viewing evidence of back surfaces) around the working surface.

Back to Gruff page