Motion processing model.

A) Schematic of a basic correlator elementary motion detector (EMD) used as the fundamental motion detection algorithm in this paper. B) Diagrammatic representation of the fully elaborated motion processing model used in this study. C) Legend describing the symbolic representations used in B). Each stage of the model represents the processing occurring on a pixel-wise basis within the insect visual system. Connections between near-by processing columns (nearest or next-nearest neighbors) in the 2D network occur between stages, mostly in the form of spatial high-pass filtering, with the only global stage a final spatial summation at the start of stage 5. Each stage is further divided into smaller processing steps involving operations such as 1st order low-pass filtering, centre-surround antagonism, non-linear gains or divisive feedback. Further detail is presented in the text.