Image Inpainting using Efficient Patch Selection Technique

Abstract

Image inpainting is a well growing research area in the field of image processing. It is used to reconstruct the spoiled images due to aging. It can also reproduce the images by removing any unwanted objects in the image. Exemplar inpainting methods gain the most attention for object removal applications. The major difficulty observed in these methods includes mismatching of patches while selecting the exemplar patches, which leads to fill by miscellaneous objects in the target region. In this paper, the new patch selection is employed to overcome the above stated issue. The combination of Sum of Squared Difference (SSD) and Logarithmic Similarity (LS) is proposed in this paper. The results generated from the proposed work are compared to the available method in the literature qualitatively. It shows that proposed work outperformed the state-of-art works.

Index Terms: Mismatch, Exemplar, SSD, Logarithmic Similarity, qualitative