Spatio-temporal image filtering with cellular neural networks

This paper describes a cellular neural network for spatio-temporal image filtering. The filter implemented is similar to the space-time Gabor filter, which has been used as a preprocessing step in several algorithms for image motion analysis. Although it appears that algorithms based upon spatio-temporal filters are quite robust, a significant drawback is that they are computationally intensive on a digital computer. The eventual goal of this research is to implement the filter described here in an analog VLSI chip along with photosensors to sense the image. This chip will simultaneously sense and process the image in parallel, relieving the bottleneck of the filter computations