Sufficient conditions for one-dimensional cellular neural networks to perform connected component detection (Q708585)

The authors consider 1D cellular neural networks (CNNs) with opposite-sign template described by the system of differential equations \[ x'_i(t)=-x_i(t)+a_{-1}y_{i-1}(t)+a_0y_i(t)+a_1y_{i+1}(t), \quad i=1,2,\dots,n \] with \(y_0(t)=y_{n+1}(t)=-1\), \(t\geq 0\), where \(x_i(t)\) and \(y_i(t)\) represent the state and output of the \(i\)-th cell at time \(t\), \(n\) is the number of cells, \(y_i(t)=\frac{1}{2}(|x_i(t)+1|-|x_i(t)-1|)\), and \(A=[a_{-1}, a_0, a_1]=[r, p, -s]\) with \(r>0\), \(p>1\), \(s>0\). Sufficient conditions are provided for such 1D CNNs to perform a connected component detection.