FindElements output: contours plus the per-component BoundingBoxes ConnectedComponents already computes on GPU as a matter of course. More...
#include <VisionInsight.hpp>
Collaboration diagram for MayaFlux::Kinesis::Vision::FindElementsAnalysis:Public Attributes | |
| std::vector< BoundingBox > | boxes |
| std::vector< Contour > | contours |
| std::vector< float > | nearest_neighbor_distance_px |
| std::vector< ElementShape > | shapes |
Detailed Description
FindElements output: contours plus the per-component BoundingBoxes ConnectedComponents already computes on GPU as a matter of course.
boxes is indexed by label 1..count (boxes[i] has label_id == i + 1), independent of contours' order (contours may be reordered by contour_topk_select.comp's area sort, or filtered by min_area). Join a contour to its own box via contour.label_id: boxes[contour.label_id - 1]. A hole contour's label_id is 0 and has no corresponding box; a hole is not itself a component.
shapes and nearest_neighbor_distance_px are both parallel to boxes (one entry per real component, holes excluded): shapes[i] is boxes[i]'s own compactness/aspect ratio, nearest_neighbor_distance_px[i] is the pixel distance from boxes[i]'s centroid to the closest other box's centroid (0 when boxes has fewer than two elements).
Definition at line 117 of file VisionInsight.hpp.
The documentation for this struct was generated from the following file:
