Perceptually aware image inpainting. (November 2018)

• Record Type: Journal Article
• Authors: Ding, Ding; Ram, Sundaresh; Rodriguez, Jeffrey J.
• Is Part Of: Pattern recognition. Volume 83(2018:Nov.)
• Page Start: 174
• Page End: 184
• Publication Date: 2018-11
• Subjects: Exemplar-based image inpainting -- Object removal -- PAMSE -- Texture synthesis -- Geometric structure propagation; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2018.05.025
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: Propose a new exemplar-based inpainting that inpaints both the geometrical structures and textures well. We apply an IQA-based patch similarity metric towards image inpainting. Propose an effective inpainting algorithm by using an adaptively growing source region. The proposed inpainting algorithm is effective for object removal from images. Abstract: Image inpainting is a process of reconstructing missing regions, or removing unwanted objects automatically by propagating intensity and texture information from surrounding parts of the image in a visually plausible manner. We propose a new exemplar-based image inpainting algorithm, which uses the recently developed metric called the perceptual-fidelity aware mean squared error (PAMSE). The PAMSE is a Gaussian-smoothed mean squared error (MSE) and approximates a weighted sum of the gradient of MSE, the Laplacian of MSE, and MSE itself. We show that, compared to MSE, PAMSE is a promising perceptual fidelity metric for application to image inpainting and leads to better performance in propagating texture and geometric structure simultaneously.