Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.


: Michaelsen, E.; Meidow, J.


Michaelsen, Eckart; Meidow, Jochen:
Hierarchical Perceptual Grouping for Object Recognition. Theoretical Views and Gestalt-Law Applications
Cham: Springer International Publishing, 2019 (Advances in Computer Vision and Pattern Recognition)
ISBN: 978-3-030-04039-0
ISBN: 978-3-030-04040-6
ISBN: 978-3-030-04041-3
Aufsatz in Buch
Fraunhofer IOSB ()

Images contain repetitions of structure. Very often, these similar things are ordered along certain laws, which are long known and referred to as Gestalt laws. Things are arranged in patterns. Moreover, often there is hierarchy in such arrangements. Small things are arranged in patterns within bigger things, and these in turn may again be part of an even greater aggregate, and so forth. This chapter introduces the analysis of such hierarchies informally by discussing a small probe of example images. Thus motivated a domain is introduced in which these things can be manipulated by machines. We call them Gestalten. They feature location, scale, orientation, and so forth. But their most important property is their assessment—a number between zero and one. This chapter also reviews related approaches to the topic. There are several important books that give alternative views, and continuous competition workshops on symmetry recognition are held along the major machine vision conferences, and of course, there are key publications in major journals.