Scale Perception and Choice of Regularization Parameter in Total Variation Regularization

I will discuss the basic problem of TV regularization in image processing, and I will discuss the two fundamental properties of TV regularization: edge-preservation and scale-dependency. I will discuss how to use TV regularization to recognize the scale of individual image features. I will then present an algorithm for automatically eliminating from an image all features whose scales are below a user-chosen scale threshold, and I will give examples of its application to several images