作者: Bernard C. Levy
DOI: 10.1007/978-1-4612-3462-3_2
关键词:
摘要: In [1], it was shown that 2-D discrete Gauss-Markov random fields can be characterized in terms of a noncausal nearest-neighbor model (NNM) driven by locally correlated noise. This result is used here to obtain simple solution the smoothing problem for fields. It smoother has structure same type as original field, and error itself field. Since operator describing dynamics positive self-adjoint, implemented using efficient iterative algorithms elliptic PDEs.