This experiment was developed to test the system's model-building capabilities, necessary for reverse engineering. Set-up was similar to the first experiment, but with the camera mounted on a robot arm to allow constrained motion between views.
The experiment started by taking images for the object from two positions,
defined by the operator. These were used to
generate two sets of contours which were fed into the stereo module for depth
estimation. Using the stereo module, an initial set of world coordinates
for these contours was generated and converted to an _1model.
A block diagram for the experiment is shown in Figure 5. The resulting model for one of the two machined parts used is shown in Figure 6.
