Silhouette Restoration

Silhouette estimation

“…A maximum-likelihood (ML) estimator is shown to work well for data corrupted with modest amounts of noise. However, a maximum a posteriori (MAP) estimator, first presented in [7], provides needed management of noise amplification when noise levels are more substantial. We also present a practical MAP estimator that accommodates misregistration between the data and the model as well as acquisition over finite fields of view.…”