Digital Signal Processing Reference
In-Depth Information
Figure 4.12.
Bootstrap Results: for a) coastguard sequence, frames 30-39, b) con-
tainer sequence 0-45, downsampled by 5 and c) hall monitor sequence, frames 30-39.
Analysis of results are in section 4.12. Each differently textured region is a separate
initial surface estimate in the frame.
ysis in our object detection. In the hall monitor sequence, the person is
detected as two different objects, because of his non-rigid motion. These
regions may also be joined together through some proximity measure,
but, once again, our system currently has no high-level models of region
interact ions.
Search WWH ::
Custom Search