An Improved and Optimized Content-Aware Resizing Algorithm for Images with Densely Situated Foreground Objects

An image in general consists of a combination of significant objects in the foreground and not-so-significant objects in the background. Content aware image resizing or seam carving is a process of resizing an image while maintaining the significant objects (the foreground) in proper visual saliency. The standard algorithms, however, often generate unpredictable distortions in images with densely situated foreground objects. The optimized content aware image resizing (OCAIR) algorithm presented herein, uses iterative graph cuts and edge detection to generate an energy map based on the important sections of the image, so that the resized image does not exhibit unpredictable artefacts. An improved energy map generation algorithm is designed here, which not only marks out the important foreground elements quicker than previously available techniques, but also uses that information to quantity the amount of distortion (if any) that might take place after adding or deleting seams by means of calculating a distortion factor. The process being considerably faster than previous algorithms, allows precise modifications to the input parameters to obtain a well-doctored final image.