5.4 Results

The first experiment, to test the algorithm, uses two images of a calibration grid. Two planes were detected (see figure 5), and the angle between the two planes was found to be 89.91 degrees which is a relative error of 0.1%.

  
Figure 5: Two automatically detected planes of a calibration grid. (angle 89.91 degrees.)

A sequence of uncalibrated PAL images were captured using a PULNIX camera. The images were projectively calibrated using a set of manual point matches, and matched edges were obtained using edge matching software by Pollard et al. [12]. The results of the algorithm described in Section 5.3 are shown in Figures 6 and 7. Figure 6 shows sets of lines which agree with planes in the world, and Figure 7 shows the two incorrect planes found by the algorithm.

A second sequence of three images of Fitzwilliam Museum were captured using a digital camera, and the results are shown in figure 8.

• 1 Introduction
• 2 Review
• 3 Describing Occlusion Junctions
• 4 Detecting occlusion junctions using three views
  • 4.1 Algorithm
  • 4.2 Results and discussion
• 5 Planarity constraint of two lines
  • 5.1 The occlusion pseudo-junction
  • 5.2 The planarity constraint
  • 5.3 Method
  • 5.4 Results
  • 5.5 Discussion
• 6 Conclusion
• References
• A Derivation of the junction point motion of two lines
• B Alternate proof of planarity constraint.
• About this document ...