Sparsity-guided saliency detection for remote sensing images

Abstract

Abstract. Traditional saliency detection can effectively detect possible objects using an attentional mechanism instead of automatic object detection, and thus is widely used in natural scene detection. However, it may fail to extract salient objects accurately from remote sensing images, which have their own characteristics such as large data volumes, multiple resolutions, illumination variation, and complex texture structure. We propose a sparsity-guided saliency detection model for remote sensing images that uses a sparse representation to obtain the high-level global and background cues for saliency map integration. Specifically, it first uses pixel-level global cues and background prior information to construct two dictionaries that are used to characterize the global and background properties of remote sensing images. It then employs a sparse representation for the high-level cues. Finally, a Bayesian formula is applied to integrate the saliency maps generated by both types of high-level cues. Experimental results on remote sensing image datasets that include various objects under complex conditions demonstrate the effectiveness and feasibility of the proposed method.

Nested Network With Two-Stream Pyramid for Salient Object Detection in Optical Remote Sensing Images

“…7. The PR curve describes [18] 0.7080 0.6268 0.6874 0.0626 0.7662 RRWR [48] 0.5782 0.6591 0.5950 0.1324 0.6835 HDCT [49] 0.6071 0.4969 0.5775 0.1309 0.6197 DSG [50] 0.6843 0.6007 0.6630 0.1041 0.7195 MILPS [51] 0.6954 0.6549 0.6856 0.0913 0.7361 RCRR [15] 0.5782 0.6552 0.5944 0.1277 0.6849 SSD [29] 0.5188 0.4066 0.4878 0.1126 0.5838 SPS [31] 0.4539 0.4154 0.4444 0.1232 0.5758 ASD [33] 0.5582 0.4049 0.5133 0.2119 0.5477 DSS [24] 0.8125 0.7014 0.7838 0.0363 0.8262 RADF [25] 0.8311 0.6724 0.7881 0.0382 0.8259 R3Net [16] 0.8386 0.6932 0.7998 0.0399 0.8141 RFCN [28] 0.8239 0.7376 0.8023 0.0293 0.8437 LV-Net 0.8672 0.7653 0.8414 0.0207 0.8815 the different combination of precision and recall scores, and the closer the PR curve is to the coordinates (1, 1), the better performance achieves. Compared with other methods, the proposed LV-Net algorithm achieves a higher recall score while achieving a higher precision score, and thus, its PR curve is much higher than other methods with a large margin.…”

A Parallel Down-Up Fusion Network for Salient Object Detection in Optical Remote Sensing Images

“…In the experiments, we compare the proposed PDF-Net with 13 state-of-the-art SOD methods on the testing subset of the ORSSD dataset, including four unsupervised methods for NSIs (i.e., RBD [44], DSG [70], MILPS [71], and RCRR [49]), five deep learning-based methods for NSIs (i.e., DSS [51], RADF [52], R3Net [13], RFCN [14], sponding ground truth images, the results of RBD [44], RCRR [49], SSD [32], RADF [52], PoolNet [15],…”

“…and PoolNet [15]), and four methods for optical RSIs (i.e., SSD [32], SPS [60], ASD [61], and LVNet [33]). For a fair comparison, the results of competitors are generated by the released codes or directly provided by the authors.Moreover, the deep learningbased methods for NSIs are retrained on the same training data of the ORSSD dataset using their default parameter settings.…”

“…For the SOD in optical RSI, there is less relevant research. Zhao et al [32] achieved SOD in optical RSIs by using the sparse representation with the help of the global and background cues, and collected two optical RSI datasets with the corresponding pixellevel saliency masks, but both datasets are not publicly available. Zhang et al [58] proposed a feature fusion model under the low-rank matrix recovery framework to achieve SOD in RSIs.…”

“…foreground annotation [4], segmentation [5,6], retargeting [7,8], quality assessment [9,10], thumbnail creation [11], and video summarization [12]. With different types of input data, SOD models can be classified into different categories, such as RGB SOD models [13,14,15,16], RGB-D SOD models for RGB-D images [17,18,19,20], co-saliency detection models for image group [21,22,23,24], video SOD for video sequences [25,26,27,28], light filed SOD models for light filed image [29,30,31], and remote sensing SOD models for remote sensing image (RSI) [32,33,34]. With the great need in military and civilian applications, the optical RSI has become more and more significant over the past few years.…”