Perceptual evaluation of digital image completion quality

Recently, image completion or inpainting has become a hot research topic because of its extensive applications in various real-world applications. Many papers have been published on this topic, but there are very few studies devoted to objective image inpainting evaluation. The intend of this work is to propose a solution based on Image Quality Metrics (IQMs). The idea is to adapt some IQMs by taking into account some specificities and constraints related to image inpainting objectives. The main characteristics used in this approach are the visual coherence of the recovered regions and the visual saliency. In this paper the notion of Inpainted Image Quality Assessment is introduced and discussed trough some experimental results. The obtained results demonstrate that the proposed image quality metric is well adapted to image inpainting evaluation.