Curves approximation by the Abel-Poisson kernel and its bit-level implementation

A linear operator for digital contour smoothing by means of the Abel-Poisson kernel is proposed. This operator is represented by a circulant Toeplitz matrix with integer coefficients which performs weakly feasible smoothing and can be considered as a suitable length estimator. Further, for this operator time-optimal bit-level systolic array is suggested. The performance of the proposed operator is also presented