Rights / licenseIn Copyright - Non-Commercial Use Permitted
This report presents an object part detection method using a par- ticle lter. The method is adapted to a range camera that provides 3D information with a high data rate. However, the data is aected by considerable measurement noise and distortion. Thus, the range data is quantized to cope more eciently with the high data vol- ume and segmented into primitive parts with morphological oper- ators to assure processing speed. Measurement noise, outliers and segmentation errors are handled with a particle lter used here as a soft decision tree to detect object parts over several frames Show more
External linksSearch via SFX
SubjectCOMPUTER VISION + SCENE UNDERSTANDING (ARTIFICIAL INTELLIGENCE); COMPUTERVISION (KÜNSTLICHE INTELLIGENZ)
Organisational unit03737 - Siegwart, Roland Y.
MoreShow all metadata