Artifact Removal and Image Inpainting Using Fuzzy Confidence Term

  • Shweta Dhondse, Milind Mushrif

Abstract

The inpainting of image is a process in which the damaged region or region which is missing is completed by diffusing information taken from the known region. Inpainting using exemplar method is in general used for filling large damaged or unknown areas left after object removal. The main components of exemplar based inpainting are to compute the filling order through priority term and finding the best match patch by iterative algorithm. Though the exemplar based approach is proved to be simple and efficient , its main drawback is rapidly dropping priority value. A novel approach is introduced in exemplar based image inpainting by extending the fuzzy sets theory  to cover  uncertainty in digital images. A modified fuzzy confidence term is proposed to control the rapidly dropping priority value. Experiments are performed on images varying in amount of texture and structure content. The  results are compared with the standard inpainting algorithm and qualitative analysis is done on the basis of more visibly plausible output.        

Published
2020-08-01
Section
Articles