Needle Map Recovery using Robust Regularizers

Philip L. Worthington Edwin R. Hancock
Department of Computer Science, University of York, UK.

Abstract:

This paper describes how robust error-kernels can be used as smoothness priors in recovering shape from shading. Conventionally, the smoothness error is added to the data-closeness (or brightness-error) as a quadratic regularizer. We introduce a novel regularizer of the form . This regularizer has a sigmoidal derivative and offers a compromise between premature outlier rejection and oversmoothing. Experiments on synthetic and real-world data reveal that this robust regularizer enhances needle-map recovery, and reduces sensitivity to noise.

 
• 1 Introduction
• 2 Shape from Shading
• 3 The Robust Regularizer
• 4 Experiments
• 5 Conclusions and Outlook
• References