AUTOMATED METHOD AND DEVICE FOR ENSURING THE PERCEPTIVE INVARIANCE OF A SPATIO-TEMPORAL EVENT DYNAMICALLY FOR EXTRACTING UNIFYED SEMANTIC REPRESENTATIONS

摘要

Method and automated device capable of ensuring the dynamic perceptual invariance of an event (90) in order to extract unified semantic representations. The event is perceived by a data transducer (130) outputting a signal (HD) referenced (x, y), transformed into a signal (MAP1) referenced (i, j) through a unit (Dec) providing a Gaussian filtering (200) parametrized by w and decimated (300) by a coefficient k, and transformed into a signal (MAP2) referenced (X, Y) representative of the invariant event through a unit (ROI). A unit (71 ') transforms the signal (MAP1) into an oriented edge signal (ImBo) and curvature (ImCb) presented to a dynamic attractor (80_0) which statistically converts this information into curvature cb0 and average bo0 orientation, center of gravity i0, j0 and size of the point cloud ap0, bp0 to feed an invariance calculation unit (Inv) which delivers the parameters w, k and the addresses (X, Y) of the signal (MAP2). The invention applies in the field of neuroscience as an electronically integrable storage unit.