Parametric Blur Estimation Using the Generalized Cross-Validation Criterion and a Smoothness Constraint on the Image

Parametric blur estimation is proposed using the Generalized Cross-Validation Criterion (GCV) and a smoothness constraint on the image. This approach bypasses the need to identify the image AR parameters. The blur synthesis domains that may be used in the GCV criterion are compared experimentally. The criterion is then applied, in a myopic scheme, on blurred and unblurred images. Results with synthetic and real blurs are discussed.