Receptive field structures for recognition
Author(s)
Balas, Benjamin; Sinha, Pawan
DownloadMIT-CSAIL-TR-2005-015.ps (33.19Mb)
Additional downloads
Metadata
Show full item recordAbstract
Localized operators, like Gabor wavelets and difference-of-Gaussian filters, are considered to be useful tools for image representation. This is due to their ability to form a Âsparse code that can serve as a basis set for high-fidelity reconstruction of natural images. However, for many visual tasks, the more appropriate criterion of representational efficacy is ÂrecognitionÂ, rather than ÂreconstructionÂ. It is unclear whether simple local features provide the stability necessary to subserve robust recognition of complex objects. In this paper, we search the space of two-lobed differential operators for those that constitute a good representational code under recognition/discrimination criteria. We find that a novel operator, which we call the Âdissociated dipole displays useful properties in this regard. We describe simple computational experiments to assess the merits of such dipoles relative to the more traditional local operators. The results suggest that non-local operators constitute a vocabulary that is stable across a range of image transformations.
Date issued
2005-03-01Other identifiers
MIT-CSAIL-TR-2005-015
AIM-2005-006
CBCL-246
Series/Report no.
Massachusetts Institute of Technology Computer Science and Artificial Intelligence Laboratory
Keywords
AI, object recognition, face recognition, sparse coding