This thesis focuses on the timely topic of spatiotemporal filtering by means of bioinspired analog VLSI array processors in the continuous-time discrete-space domain. A novel extension to the seminal Cellular-Neural-Network (CNN) filtering array architecture, based on 'time-derivative diffusion' between neighboring cells, is proposed. This results in a paradigm with rich spatiotemporal filtering dynamics. Apart from the wider range ofspatiotemporal filtering functions possible with the proposed extension, its practical advantage lies in the intimate one-to-one relationship between rational spatiotemporal transfer function polynomial terms and the network connections. More specifically, in this thesis:• The new 'time-derivative-diffusion' network paradigm is formally presented and its potential investigated by means of filtering examples and analysis tool? based on linear systems theory. Based on this, the ability of the proposed network to produce novel space-time nonseparable filtering operations is investigated. • A generic intuitive filter synthesis procedure based on 3D spatiotemporal pole placement is presented with 3D spatiotemporal bandpass filters as design examples.