A robust salient object detection using edge enhanced global topographical saliency

Complex salient object detection is the most challenging task in clutter background images. In this prevailing problem, global contrast-based methods are comprehensively preferred. But these methods fail in preserving the structure, shape and broader related geometrical information. Aiming at these limitations, the proposed method uses global contrast and iterative Laplacian of Gaussian to generate initial global topographical saliency. In this topographical saliency, iterative Laplacian of Gaussian is used to preserve the structural, shape and broader related geometrical information. This global topographical saliency is used as a reference plane for integrating regional saliencies. The color, spatial and distance based regional saliencies are integrated into the boundary enhanced global topographical saliency to improve the substantial information of the object. Boundary-based Gaussian weighted, background suppression model, is used to remove the background and edge-effects. Finally, central saliency addition is used to enhance the final saliency. The proposed method is compared with recent six global contrasts based state-of-art methods, two deep learning based methods and four publicly available datasets. The experimental result presented here shows that the proposed method performs better in comparison to the state-of-the-art methods.