| Abstract: | In recent years, content-aware image retargeting has been an active research topic with the rapid growth of mobile devices. Retargeting techniques usually begin with computing an importance map, which represents the image regionsthat draw the human attention. Saliency map is highly considered as a vital component of the importance map to preserve the attended regions of the input image after resizing. Context-aware saliency detection has recently emerged as a promising direction in producing accurate maps. However, its high computational complexity, severely, limits its use in practical applications. In this paper, we proposed a local search window to limit the search space, when looking for similar patches within the image. Experimental study over a set of 86 benchmark images showedthat the proposed approach significantly reduces the running time, while,at the same time, produces accurate results compared to the original Context-aware saliency detection algorithm. Moreover, we conducted a set of experiments to empirically set the size of the proposed search window
|