Parameterized Cellular Automata in Image Segmentation

This paper investigates a novel update rule for multi-state Cellular Automata (CA) in the context of greyscale image segmentation. The update rule is parameterized and takes into account the features of neighbouring cells compared to the features of the current cell. We use the resulting CA to segment several real-world images. During this process we also study the influence of the rule parameters and neighbourhood scheme using different evaluation measures.