Improved exemplar-based image inpainting approach

Abstract

Image inpainting is a very common technique which is widely used to bring back or recover an image when it gets destroyed, or when there is an intention to perform a morphological operation on an image. It is a simple process to fill up the pixels of a particular region. The task is very similar to that of a skilled painter who has to draw or remove an object from its painting on its canvas. Inpainting’s application may vary from its usage which includes object removal, object replacement etc. The aim of this process is to change the image properties by removing or adding objects. It has been also used to recover lost pixels or block of pixels of an image during a transmission of image through a noisy channel. It has been a useful method for red-eye removal and default stamped date from photographs clicked through primitive cameras. In this paper, we have built an exemplar-based image inpainting tool using basic functions of MATLAB. We have discussed the results by comparing the output generated by the image inpainting tool with Adobe Photoshop results to compare the results and check how much efficient the tool is.