An improved content aware image resizing algorithm based on a novel adaptive seam detection technique

An image can be considered to be a combination of both significant (foreground) objects and some less significant (background) objects. Content aware image resizing (CAIR) algorithm uses the different edge detection methods to segregate the useful objects from the background. When applied to an image, CAIR can resize the image to a very different aspect ratio without destroying the aspect ratio of the useful objects in the image. However, this method fails when the useful objects in the image are very closely situated. To take care of this, this paper proposes and develops a modified version of the algorithm. Instead of merely finding the edges, the important objects are detected by drawing contours around them with the help of level set based Chan Vese Image Segmentation algorithm and constant convergence rate Modified Delta Bar Delta learning algorithm. Then Seam Carving algorithm is applied which uses Dynamic Programming. A seam which is a 8 connected curved path from top to bottom (vertical seam) or left to right (horizontal seam) is drawn on the unnoticeable pixels of the lesser significant portions by the process of seam carving which helps to resize the image to a new size. The optimum path of the seam is defined by an image energy function which protects the content of the image. If the seams are continuously removed and inserted then the size of an image can be expanded and contracted respectively in both directions.