Background-based Saliency Research Articles

Salient object detection via cross diffusion-based compactness on multiple graphs

The diffusion-based graph has been widely used in saliency detection. Most of the existing methods treat the image boundary patches as background seeds, which may result in the imprecise saliency map if the salient region touches the image boundaries. In this paper, we propose a salient object detection approach via cross diffusion-based compactness on multiple graphs. Firstly, we extract multi-view image features including low-level image features, mid-level image cues (low-level saliency priors), and background-based saliency map. Then, we compute the respective similarity matrix to construct the corresponding graph, and a cross-diffusion algorithm is presented that diffuses each similarity matrix on other’s graphs rather than at itself graph, helping to the compactness-based saliency maps. Additionally, for well propagating the saliency values, we model a propagation mechanism based on cellular automata, by linearly incorporating low-level image features and mid-level image cues together to generate a superior impact factor matrix. Extensive experiment results demonstrate that the proposed method achieves better saliency detection performance against the unsupervised state-of-the-art methods on three public datasets.

Salient object detection via two-stage absorbing Markov chain based on background and foreground

This paper proposes a saliency detection method via two-stage absorbing Markov chain based on background and foreground for detecting salient objects in images. Firstly, image preprocessing is performed, followed by convex hull construction and superpixel segmentation, to prepare for subsequent processing. Secondly, according to the boundary connectivity, the superpixels with lower background probability value in the candidate boundary background set B0 are deleted, and the boundary background set B1 is obtained. With the saliency values of the nodes in the boundary-prior saliency map Sbg1, the background seeds are added appropriately in the region outside the candidate boundary background set B0 and the convex hull H, and the background seed set B is obtained after update. Then, the background-absorbing Markov chain is constructed to generate background-absorbing saliency map Sbg2. By fusing the saliency maps Sbg1 and Sbg2, the first-stage background-based saliency map Sbg is obtained. Thirdly, in the range of the convex hull H, the foreground seed set F is determined according to the saliency map Sbg. Then, the foreground-absorbing Markov chain is constructed, to obtain the second-stage foreground-absorbing saliency map Sfg. Finally, the saliency maps Sbg and Sfg of the two stages are combined to obtain a fused saliency map S, and the final saliency map S∗ is obtained after optimization through smoothing mechanism. Compared with the traditional methods, the performance of the proposed method is significantly improved. The proposed method is tested on three public image datasets, and it shows great accuracy in detecting salient objects.

Integrating manifold ranking with boundary expansion and corners clustering for saliency detection of home scene

In this paper, we propose a novel framework for saliency detection of home scene by exploiting manifold ranking, boundary expansion, and corners clustering. Our proposed method firstly combines color cues in RGB and CIELab to select image boundary seeds, and exclude the ones which might be located at salient objects as much as possible. Then, we utilize the boundary seeds on each image boundary as the queries of manifold ranking to compute saliency and integrate them for a background-based saliency map. For the foreground-based saliency detection. Boundary expansion combined with background-based saliency map highlights foreground regions, which are regarded as queries for a foreground-based saliency map. Moreover, we achieve center prior saliency map through multi-scale Harris corner detection and corners clustering to further highlight salient regions and suppress background regions. Finally, we integrate the three saliency maps via the proposed unified framework for a more accurate and smooth saliency map. Both qualitative and quantitative experimental results indicate that our proposed method can deliver better performance than several state-of-the-art saliency detection methods as a whole.

Local Regression Ranking for Saliency Detection.

Saliency detection is an important and challenging research topic due to the variety and complex of the background and saliency regions. In this paper, we present a novel unsupervised saliency detection approach by exploiting a learning-based ranking framework. First, the local linear regression model is adopted to simulate the local manifold structure of every image element, which is approximately linear. Using the background queries from the boundary prior, we construct a unified objective function to globally minimize all the errors of the local models for the whole image element points. The Laplacian matrix is learned via optimizing the unified objective function. Low-level image features as well as high-level semantic information extracted from deep neural networks are used for the Laplacian matrix learning. Based on the learnt Laplacian matrix, the saliency of the image element is measured as the relevance ranking to the background queries. The foreground queries are obtained from the background-based saliency and the relevance ranking to the foreground queries is calculated in the same way as the background-based saliency. Second, we calculate an enhanced similarity matrix by fusing two different-level deep feature metrics through cross diffusion. A propagation algorithm uses this enhanced similarity matrix to better exploit the intrinsic relevance of similar regions and improve the saliency ranking results effectively. Results on four benchmark datasets with pixel-wise accurate labelling demonstrate that the proposed unsupervised method shows better performance compared with the newest state-of-the-art methods and is competitive with deep learning-based methods.

Saliency detection via background prior and foreground seeds

As an image pre-processing technology, saliency detection (DS) can be used in a wide variety of visual tasks. A bottom-up method of DS via background prior and foreground seeds is proposed. To highlight the object and suppress the background noise, two saliency maps are obtained by using the background prior and foreground seeds. First, we use global colour and spatial distance matrices to compute a background saliency map. To evenly emphasize the saliency region, the single-layer cellular automata is used to refine the background-based saliency map. Second, a set of foreground seeds is obtained from the refined background-based saliency map. Then, the foreground-based saliency map is calculated based on the foreground seeds and refined by biased Gaussian filtering. The proposed method could emphasize the foreground target, as well as restrain the background noise. The experiment results show that our method has good ability in saliency detection.

Robust pixelwise saliency detection via progressive graph rankings

In this paper, we propose a novel saliency detection method based on superpixel-to-pixel level optimization. First, we segment the input image into superpixels under four scales. For each scale, we construct a k-regular basic graph with these superpixels as nodes. Furthermore, we enlarge the basic graph with virtual absorbing nodes and utilize absorbing Markov chain ranking to calculate background-based saliency. Second, for each scale, we obtain robust foreground queries from the previous result, and use manifold ranking to obtain foreground-based saliency. Third, a regularized random walk ranking based on the pixelwise graph for each scale is used to diffuse the saliency values among pixels. Finally, we obtain four saliency maps for the input image and integrate them together for the final saliency map. Extensive experiments on several challenging datasets reveal that the proposed method performs better in terms of precision, recall and F-measure values. Despite complex backgrounds, our method performs better in detecting small and/or multiple salient objects than other state-of-the-art methods as a whole.

Combining multi-layer integration algorithm with background prior and label propagation for saliency detection

In this paper, we propose a novel approach to automatically detect salient regions in an image. Firstly, some corner superpixels serve as the background labels and the saliency of other superpixels are determined by ranking their similarities to the background labels based on ranking algorithm. Subsequently, we further employ an objectness measure to pick out and propagate foreground labels. Furthermore, an integration algorithm is devised to fuse both background-based saliency map and foreground-based saliency map, meanwhile an original energy function is acted as refinement before integration. Finally, results from multiscale saliency maps are integrated to further improve the detection performance. Our experimental results on five benchmark datasets demonstrate the effectiveness of the proposed method. Our method produces more accurate saliency maps with better precision-recall curve, higher F-measure and lower mean absolute error than other 13 state-of-the-arts approaches on ASD, SED, ECSSD, iCoSeg and PASCAL-S datasets.

Optimizing multi-graph learning based salient object detection

In this paper, we propose a novel bottom-up saliency detection algorithm to effectively detect salient objects. Different from most existing methods that are not robust to complex scenes, we utilize multi-graph learning to take various scenes into consideration. First, multiple features are used to represent superpixels, and then measured by multiple distance metrics to construct multiple graphs. The motivation is to take advantage of their complementary information to cope with different environments. Second, fixation and boundary cues are respectively used as foreground and background seeds. The fixation is effective for crowded backgrounds because of the observation that regions within eye fixations are very likely the image foreground. Third, we integrate the multiple graphs and seeds into a regularized and optimized multi-graph based learning framework to effectively generate foreground-based and background-based saliency maps. Finally, we integrate these two saliency maps to obtain a more smooth and accurate saliency map. Extensive experiments are conducted on five benchmark datasets. Experimental results show that the proposed bottom-up saliency detection method yields comparable or better results against the state-of-the-art methods, and is robust to both cluttered and clean scenes.

Salient object detection via color and texture cues

In this paper, we present a new bottom-up salient object detection approach by constructing two graphs using color and texture features within the manifold ranking framework. First, we calculate the saliency of boundary patches and exclude the ones with high saliency which might be a part of saliency object. Second, we adopt a two-stage scheme for salient detection via affinity propagation clustering and graph-based manifold ranking. The background-based saliency detection aims to obtain the salient object regions as much as possible. In the foreground-based saliency detection, a similar computation is processed as that in the former step and yet slightly different. Instead of simultaneously using all the extracted boundary patches or foreground patches as queries, we compute saliency by using the patches in each cluster in turn and integrating them. At last, the final saliency map is generated by linearly combining two saliency maps respectively exploring color and texture cues. Both qualitative and quantitative evaluations on three publicly available datasets demonstrate the robustness and efficiency of our proposed approach against 21 state-of-the-art methods.

Saliency detection integrating both background and foreground information

In this paper, we propose a novel saliency detection algorithm. The saliency of an image element is defined not only as its contrast to the background but as its similarity to the foreground. First, we extract background seeds as well as their spatial layout information from image boundaries to compute the background-based saliency map. Second, we generate a compact foreground region from the first-stage saliency map to describe the appearance and location of the salient object and calculate the foreground-based saliency map accordingly. We integrate these two saliency maps and further refine the unified one to obtain a more smooth and accurate saliency map. Each component of the presented algorithm is evaluated on the public available datasets and the experimental results also show that the presented algorithm achieves favorable performance compared to the state-of-the-art methods.

Saliency detection via background and foreground seed selection

In this paper, we propose a bottom-up visual saliency detection algorithm. Different from most previous methods that mainly concentrate on image object, we take both background and foreground into consideration. First, we collect background seeds from image border superpixels by boundary information and calculate a background-based saliency map. Second, we select foreground seeds by segmenting the first-stage saliency map via adaptive threshold and compute a foreground-based saliency map. Third, the two saliency maps are integrated by the proposed unified function. Finally, we refine the integrated result to obtain a more smooth and accurate saliency map. Moreover, the unified formula also proves to be effective in combining the proposed approach with other models. Experiments on publicly available data sets demonstrate that the proposed algorithm performs favorably against the state-of-the-art methods.