The class of digital Gabor filters, which plays an important role in image processing, is substantiated on spatial filtering characteristics of certain cells in the mammal brain. However, the way these filters are implemented in technical systems is mostly not based on biological backgrounds. Here, a novel design of a digital Gabor filter bank is proposed. In contrast to known implementations of Gabor filters, the presented architecture is biologically motivated and therefore consists of a cascade of sub-filters covering the major processing layers found in mammal visual systems - from photoreceptors over ganglion cells in the retina up to simple cells in the primary visual cortex. It is also shown; that this cascade provides several advantages compared to known implementation techniques, e. g. the calculation of valuable inter-signals, further possibilities for parallel realizations, and a noticeable reduction of computation time. Unlike conventional image processing systems, the presented filter bank is based upon a hexagonal sampling scheme - like in biology. Thus, fewer samples are needed compared to a rectangular sampling while complying with SHANNON, that reduces the computation time for processing. However, other features of the hexagonal grid, like superior symmetry and definite neighborhood are advantageous in further image processing aspects as well.