Evolution of Image Inpainting: From Image Processing Based Methods to Learning Based Method
摘要
Image inpainting is the process of recovering damaged areas in images in an undetectable manner. It has numerous applications, including the restoration of damaged images, paintings, and movies, as well as the removal of selected objects such as text, lines, subtitles, publicity, and stamps. The primary objective of inpainting is to reconstruct the missing region so seamlessly that observers cannot detect any manipulation. In the realm of image processing, inpainting is regarded as one of the most difficult tasks since it necessitates a thorough comprehension of the texture and structure of the image. This paper presents a survey of various image inpainting techniques, summarizing and comparing them to highlight the merits and demerits of each approach. This comparison can help researchers assess the methods they have suggested.