In this regard, Efros and Bertalmio are considered the pioneers (Paragios et al., 2006, Bertozzi et al., 2006, Wang et al., 2010) in this field and for advancing the research in texture synthesis and pixel interpolation respectively. During the reconstruction of disconnected pixels, the inpainting method uses known-information to fill unknown regions (disconnected pixels). These damaged portions/areas of an image are a set of unconnected pixels surrounded by a set of known adjacent pixels. In a computer vision and graphics context, inpainting is a method that interpolates neighbouring pixels to reconstruct damaged, or defective, portions of an image without any noticeable change on the restored regions when visually compared with the rest of the image. As a result image inpainting (henceforth inpainting) has become a state-of-the-art restoration technique. The evolution of computers in the 20th century, its frequent daily use and the development of digital tools with image manipulation capability, has encouraged users to appreciate image editing, e.g. restoration, and the application of on-screen visual display and special effects to images. Fig. 1 shows inpainting performed by hand. Image inpainting originated from an ancient technique performed by artists to restore damaged paintings or photographs with small defects such as scratches, cracks, dust and spots to maintain its quality to as close to the original as possible. To address the challenges raised from our findings, we outline some potential future works. We detail the strengths and weaknesses of each to provide new insights in the field. Convolutional Neural Networks and Generative Adversarial Networks. Then we review the deep learning methods, i.e. Exemplar-based texture synthesis, Exemplar-based structure synthesis, Diffusion-based methods, Sparse representation methods and Hybrid methods. For traditional methods, we divide the techniques into five sub-categories, i.e. To increase the clarity of our review, we use a hierarchical representation for the past state-of-the-art traditional methods and the present state-of-the-art deep learning methods. This paper presents a comprehensive review of image inpainting methods over the past decade and the commonly used performance metrics and datasets. The advent of the digital age has seen the rapid shift image storage technologies, from hard-copies to digitalised units in a less burdensome manner with the application of digital tools. Historically, these were restored by hand to maintain image quality using a process known as inpainting. However, for images on photographic material, images can have defects at the point of captured, become damaged, or degrade over time. Images can be described as visual representations or likeness of something (person or object) which can be reproduced or captured, e.g.
0 Comments
Leave a Reply. |