Discrete-time cellular neural network for thinning: A compound synthesis

The authors depict a novel thinning method by using a discrete-time cellular neural network. By extracting the inherent topographical properties, namely the midpoints of every line segment of the line image, thinning is shown to be attainable without a clumsy 7-neighbor test and stopping condition check. Circuit simplication is achieved by eliminating the usage of time-variant templates, and thus the only control signal in the network is the system clock. Overall performance of the thinning system is evaluated and is compared with previous studies