Article Preview
Top1. Introduction
“Inpainting is the process of reconstructing lost or deteriorated parts of images and videos. In the digital world, inpainting (also known as image interpolation or video interpolation) refers to the application of sophisticated algorithms to replace lost or corrupted parts of the image data (mainly small regions or to remove little defects” (Zhu et al., 2018; Cheng & Li, 2019; Wang et al., 2017). Image inpainting can be deployed for numerous applications; in photography and cinema for “restoration”, for eliminating the effects such as dust spots, scratches from images (known as deterioration) (Chen et al., 2016). It is also exploited for eliminating a specific object from an image or for the removal of red-eye. Image de-noising is a well-known issue faced by image processing schemes. Image denoising and image inpainting (Berntsson & Baravdish, 2014; Jiao et al., 2019; Alilou & Yaghmaee, 2017) are not similar. Image inpainting includes a broad application like safeguarding valuable old photos; chronological relics and it can also be used for video compression and error removal in video communication.
Image inpainting (Ding et al., 2018; Ign´acio & Jung, 2007) techniques are categorized into exemplar and search-oriented scheme, texture synthesis model, PDE approach, and hybrid inpainting and semi-automatic inpainting (Barbu, 2016; Muddala et al., 2016; Bhavsar & Rajagopalan, 2012 ; Tran and Tran, 2019 ; Zhang, et al., 2019 ; Cai and Wei, 2020 ; Nortje, et al., 2021). Nowadays, people require a system to retrieve damaged or scratched artwork, drawings, designs, photographs, etc. The damages might be owing to a variety of causes like scaled image, overlaid graphics or text, scratches, etc. In recent days, well-developed photo-editing equipment is obtainable for drawing, retouching, and eliminating objects using clippers from images (Dhiyanesh & Sathiyapriya, 2012; Wang et al., 2013; Li & Wen, 2012 ; Malhotra, et al., 2020 ; Parisi, et al., 2017). However, recreating the damaged region or filling up the missing information in an image remains a complex task. This system could develop and return a nice-looking image by means of image inpainting (Chunhong et al., 2017; Hu et al., 2019; Liu et al., 2019). This technique exploits the mask image and original image as input. The machine learning is found to be promising in different application (Elezaj, et al., 2021 ; Manne and Kantheti, 2021 ; kumar, 2020 ; Shirsat, 2020).