Bounding Box Detection

After the localization of the simmetry, the width of the symmetrical region is checked for the presence of two corners representing the bottom of the bounding box around the vehicle. Perspective constraints as well as size constraints are used to reduce the search. Figure 32 shows possible and impossible bottom parts of the bounding box, while figure 33 presents the results of the lower corners detection.

Figure 32: (a) correct lower part of bounding box, taking into consideration perspective constraints and knowledge on the acquisition system setup, as well as typical vehicles' size; (b) incorrect bounding boxes.

Figure 33: Detection of lower part of bounding box: (a) original with symmetry axis; (b) edges; (c) localization of the two lower corners.

This process is followed by the detection of the top part of the bounding box, which is looked for in a specific region whose location is again determined by perspective and size constraints. Figure 35 shows the search area.

Figure 35: The search area for the upper part of the bounding box is shown in dark gray. It takes into account knowledge about the typical vehicles' aspect ratio.