Total Variation Based Image Restoration with Free Local Constraints

The two main plagues of image restoration are oscillations and smoothing. Traditional image restoration techniques prevent parasitic oscillations by resorting to smooth regularization. Hence, singular image features and oscillatory textures cannot be restored. The usefulness of images obtained by smooth regularization is very limited. Regularized faces and characters lose features and are unrecognizably distorted. Oscillatory patterns and textures are not allowed by regularization methods. Smoothly regularized images are of no use in criminal/civil investigations and as court evidence. [l]. A new Total Variation based approach was developed in [2] to overcome the basic limitations of all smooth regularization algorithms. The TV-based technique use the L' norm of the magnitude of a gradient, thus making discontinuous and nonsmooth solutions possible (if they belong to the space of functions of a bounded total variation). In TV image restoration [3], the solution is obtained by sol...

• Type: Conference
• Year: 1994
• Where: ICIP
• Authors: Leonid I. Rudin, Stanley Osher
• Keywords: ICIP 1994, Image Processing, Image Restoration, Level Sets, Smooth Regularization