Motion-based segmentation for occupant tracking using a hausdorff distance heuristic

摘要

A segmentation system (16) is disclosed that allows a segmented image (31) of a vehicle occupant (18) to be identified within an overall image (the "ambient image") (38) of the area (21) that includes the image of the occupant (18). The segmented image (31) from a past sensor measurement within can help determine a region of interest (52) within the most recently captured ambient image (38). To further reduce processing time, the system (16) can be configured to assume that the bottom of segmented image (31) does not move. Differences between the various ambient images (38) captured by the sensor (22) can be used to identify movement by the occupant (18), and thus the boundary of the segmented image (31). A template image is then fitted to the boundary of the segmented image (31) for an entire range of predetermined angles. The validity of each fit within the range of angles can be evaluated. The template image can also be modified for future ambient images.