Saliency Detection for Panoramic Landscape Images of Outdoor Scenes

Byeong-Ju Han
bjhan@unist.ac.kr
UNIST

Jae-Young Sim
jysim@unist.ac.kr
UNIST

Abstract

Saliency detection has been researched for conventional images with standard aspect ratios, however, it is a challenging problem for panoramic images with wide elds of view. In this paper, we propose a saliency detection algorithm for panoramic landscape images of outdoor scenes. We observe that a typical panoramic image includes several homogeneous background regions yielding horizontally elongated distributions, as well as multiple foreground objects with arbitrary locations. We rst estimate the background of panoramic images by selecting homogeneous superpixels using geodesic similarity and analyzing their spatial distributions. Then we iteratively refine an initial saliency map derived from background estimation by computing the feature contrast only within local surrounding area whose range and shape are changed adaptively. Experimental results demonstrate that the proposed algorithm detects multiple salient objects faithfully while suppressing the background successfully, and it yields a significantly better performance of panorama saliency detection compared with the recent state-of-the-art techniques.

Experimental Results


Figure 1: Qualitative comparison of saliency detection algorithms on three panoramic images with larger aspect ratios. Figures in parenthesis refer aspect ratios. From top to bottom, we show input panoramic images, the resulting saliency maps obtained by BSCA [14], RRWR [16], SF [33], MR [15], GS [12], SO [13], and the initial (Prop.-I) and fi nal (Prop.-F) saliency maps of the proposed algorithm, respectively. The last row shows the ground truth (GT) saliency maps.


Figure 2: Qualitative comparison of saliency detection algorithms on six panoramic images. Figures in parenthesis refer aspect ratios. From top to bottom, we show input panoramic images, the resulting saliency maps obtained by BSCA [14], RRWR [16], SF [33], MR [15], GS [12], SO [13], and the initial (Prop.-I) and fi nal (Prop.-F) saliency maps of the proposed algorithm, respectively. The last row shows the ground truth (GT) saliency maps.

Publication

Byeong-Ju Han and Jae-Young Sim, “Saliency Detection for Panoramic Landscape Images of Outdoor Scenes,” Journal of Visual Communication and Image Representation, vol. 49, pp. 27-37, Nov. 2017.