Image resizing with minimum distortion

Abstract

Displays became cheap and were combined with many devices, like camera, mobile, and so on…, so there has been an increased interest on resizing methods to make the image suitable and fill any screen size. Common and known methods like cropping or resampling can cause undesirable effects such as: losses in information or distortion in perception. Recently, content-aware image resizing methods have been proposed to get rid of these problems and produce exceptional results. Seam-carving produced by Avidan and Shamir has gained attention as an effective solution. This paper discussed about this method and used it to resize (minimize and maximize) four colored images vertically and horizontally respectively, and maintained the main features of the images by deleting or repeating only the uninfluenced features. The energy map was calculated that described the basic and influential details of the image using energy function. But instead of gradient function (as in Avidan and Shamir) entropy function was used to compute the energy of the images. A vertical or a horizontal seam of pixels with minimum energy values was either deleted or inserted to resize the image. Good results were obtained especially when the image contains spaces within its details. The work was programmed using Matlab2018a.