Extracting Compact Objects Using Linked Pyramids
MARYLAND UNIV COLLEGE PARK COMPUTER VISION LAB
Pagination or Media Count:
Compact objects of arbitrary size are extracted from images using a combination of three-pyramid-based representations of image features. A gray-scale linked pyramid is used to smooth the image into uniform regions. A surroundedness pyramid is used to identify regions of interest, and a linked edge pyramid is used to delimit the boundaries of the compact objects.