Size invariance by dynamic scaling in neural vision systems.
Gotz Meierfrankenfeld, Klaus Kopecz · 1997
. Vision systems in complex environments are faced with the problem of analyzing visual information on multiple scales to support segmentation and size invariant object recognition. We propose a system which detects relevant spatial scales in images and constructs an explicit size invariant representation. Further, it provides a means of navigating through scale space in the presence of ambiguous scale information, thus adds a behaving component to image analysis. The architecture is based on biologically realistic neural networks like neural fields and neurons with bandpass receptive field characteristics. 1. Introduction The human nervous system shows an intriguing generalization performance with respect to size in object recognition tasks. Size invariant recognition of objects does not require to view the object in several sizes during training. One view of a certain size may be enough to recognize the same object presented at a different distance. Thus, scale invariant recognitio...