Generation of patterns with predefined statistical properties using Cellular Neural Networks

The following paper presents an application of Cellular Neural Networks for spontaneous generation of patterns with predefined statistical properties. The considered pattern statistical characteristics included coarseness, directionality and edge sharpness distribution. A purpose of this choice was to verify CNN capabilities in generating images that comprise structural elements significantly larger than a range of cell interconnections. Resulting CNN templates, derived using the SQP optimization algorithm, were capable of meeting majority of formulated design requirements.