Event
Researchers Propose Moving Object Segmentation Framework Based on OWL Functions
1 reports · 1 sources
Daniel Raviv and Juan D. Yepes propose an analytical framework based on OWL perception functions for segmenting multiple independently moving rigid bodies during a single camera motion. The framework encodes perceived looming L and perceived rotation ω as closed-form representations of relative 3D structure, and on this basis analytically recovers the relative translational unit direction pixel by pixel; points on the same non-rotating rigid body share the same direction, and clustering completes the segmentation. The researchers state that the method is derived in 3D using quaternions, possesses parallelism and invariance to camera rotation, and requires no training data, depth information, or semantic priors.
Report timeline
- A Framework for Segmenting Moving Objects During Camera Motion Based on the OWL Function
Researchers propose an analytical framework based on the OWL perception function for segmenting multiple independently moving rigid bodies during single-camera motion. OWL encodes perceived looming L and perceived rotation ω into a closed-form representation of relative 3D structure, from which the unit direction of relative translation can be analytically recovered per pixel from L and ω alone; points on the same non-rotating rigid body share the same direction, so clustering completes the segmentation. The method is derived in 3D using quaternions, is parallel and camera rotation-invariant, and requires no training data, depth information, or semantic priors.
arXiv cs.RO ·