On the plausibility of the discriminant center-surround hypothesis for visual saliency.

By Dashan Gao, Vijay Mahadevan, Nuno Vasconcelos. Journal of Vision, 2008

312 citations recorded by the source corpus.

It has been suggested that saliency mechanisms play a role in perceptual organization. This work evaluates the plausibility of a recently proposed generic principle for visual saliency: that all saliency decisions are optimal in a decision-theoretic sense. The discriminant saliency hypothesis is combined with the classical assumption that bottom-up saliency is a center-surround process to derive a (decision-theoretic) optimal saliency architecture. Under this architecture, the saliency of each image location is equated to the discriminant power of a set of features with respect to the classification problem that opposes stimuli at center and surround. The optimal saliency detector is derived for various stimulus modalities, including intensity, color, orientation, and motion, and shown to make accurate quantitative predictions of various psychophysics of human saliency for both static and motion stimuli. These include some classical nonlinearities of orientation and motion saliency and a Weber law that governs various types of saliency asymmetries. The discriminant saliency detectors are also applied to various saliency problems of interest in computer vision, including the prediction of human eye fixations on natural scenes, motion-based saliency in the presence of ego-motion, and background subtraction in highly dynamic scenes. In all cases, the discriminant saliency detectors outperform previously proposed methods from both the saliency and the general computer vision literatures.

View Paper: https://www.semanticscholar.org/paper/5afb01354798707950112bb4cde06ef2e4983819

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

A Benchmark of Computational Models of Saliency to Predict Human Fixations (2012, methodology): “[87] use the area under the ROC curve and KL-divergence to measure performance and compare against methods by [35], [6] and [22].”
Saliency estimation using a non-parametric low-level vision model (2011, background): “[4] approached saliency at a location as the discriminatory power of a set of features describing that location to distinguish between the region and its surround.”
Visual Saliency Based on Scale-Space Analysis in the Frequency Domain (2013, unknown): “F”
Predicting the Driver's Focus of Attention: The DR(eye)VE Project (2017, unknown): “This can be achieved by either merging saliency maps at different levels of scale and abstraction [23], or including a-priori cues about relevant objects for the task at hand [14], [21], [63].”
Learning to segment moving objects in videos (2014, methodology): “…), an image only CNN ( imgCNN ), a ﬂow only CNN ( ﬂowCNN ), our implementation of a standard center-surround saliency measure from optical ﬂow magnitude ( center-surround ) [14], and an ob-jectness detector using the 7000 category detector from the Large Scale Domain Adaptation (LSDA) work of [20].”
Spatio-Temporal Moving Object Proposals (2014, methodology): “…CNN (piCNN-class), an image only CNN (imgCNN), a ﬂow only CNN (ﬂowCNN), our implementation of a standard center-surround saliency measure from optical ﬂow magnitude [16], and an objectness detector us-ing the 7000 category detector from the Large Scale Domain Adaptation (LSDA) work of [20].”
Cracking BING and Beyond (2014, methodology): “Many literatures on saliency detection have found that object rectangles of interest prefer close to image centers [11][3][14][15], and their sizes are usually not too large or too small.”
Anomaly Detection and Localization in Crowded Scenes (2014, background): “The discriminant saliency principle can be applied to many features [31].”
Leveraging stereopsis for saliency analysis (2012, background): “A rich literature exists on image saliency analysis in cognitive psychology [36,40], neuroscience [17, 35], and vision perception [9, 29].”
Anomaly detection in crowded scenes (2010, background): “Discriminant saliency can be combined with many features [11].”
Two-Stage Learning to Predict Human Eye Fixations via SDAEs (2016, background): “[34] extended Itti’s work by modeling a discriminant center-surround hypothesis which assumes salient locations enable the discrimination between center”
Static and space-time visual saliency detection by self-resemblance. (2009, background): “Mahadevan and Vasconcelos (2008) applied this bottom-up saliency to background subtraction in highly dynamic scenes.”