Original title: Joint covariance property under geometric image transformations for spatio-temporal receptive fields according to the generalized Gaussian derivative model for visual receptive fields
Authors: Tony Lindeberg
The paper delves into how visual systems handle changes in images. Understanding these alterations is vital for both computer and biological vision. This study focuses on covariance properties related to geometric transformations in the initial layers of visual processing. They define and demonstrate a joint covariance property under various image transformations like scaling and temporal changes. This property helps characterize how these transformations work together. By uncovering how receptive field parameters need to adapt to match spatio-temporal transformations, the study sheds light on how the brain and machines process visual information amidst different image alterations.
Original article: https://arxiv.org/abs/2311.10543