Separable time-causal and time-recursive spatio-temporal receptive fields
This work addresses the need for biologically plausible and computationally efficient spatio-temporal models in computer vision and neuroscience, representing an incremental improvement over previous scale-space formulations.
The paper tackles the problem of modeling spatio-temporal receptive fields by proposing an improved model that combines Gaussian spatial filters with cascaded first-order integrators or truncated exponential filters for temporal processing. The result is a time-causal and time-recursive framework that ensures non-creation of new local extrema or zero-crossings with increasing temporal scale, with extensions for parameterizing intermediate temporal scales and enabling discrete recursive implementations.
We present an improved model and theory for time-causal and time-recursive spatio-temporal receptive fields, obtained by a combination of Gaussian receptive fields over the spatial domain and first-order integrators or equivalently truncated exponential filters coupled in cascade over the temporal domain. Compared to previous spatio-temporal scale-space formulations in terms of non-enhancement of local extrema or scale invariance, these receptive fields are based on different scale-space axiomatics over time by ensuring non-creation of new local extrema or zero-crossings with increasing temporal scale. Specifically, extensions are presented about parameterizing the intermediate temporal scale levels, analysing the resulting temporal dynamics and transferring the theory to a discrete implementation in terms of recursive filters over time.