#### DMCA

## Multiscale Discriminant Saliency for Visual Attention

### Citations

1745 | A model of saliency-based visual attention for rapid scene analysis
- Itti, Koch, et al.
- 1998
(Show Context)
Citation Context ...ry (FIT) [34] which elaborates computational principles of saliency map generation with center-surround operators and basic image features such as intensity, orientation and colors. Then, Itti et al. =-=[21]-=- implemented and released the first complete computer algorithms of FIT theory 5. Feature Integration Theory are widely accepted as principles behind visual attention partly due to its utilization of ... |

799 |
Cognitive Psychology.
- NEISSER
- 1967
(Show Context)
Citation Context ...ing scenic information. Robustness and efficiency of biological devices, the eyes and their control systems, visual paths in the brain have amazed scientists and engineers for centuries. From Neisser =-=[26]-=- to Marr [25], researchers have put intesive effort in discovering attention principles and engineering artificial systems with equivalent capability. For last two decades, this research field is domi... |

340 |
Modeling the role of salience in the allocation of overt visual attention
- Parkhurst, Law, et al.
- 2002
(Show Context)
Citation Context .... In signal processing, scale-space and spectral space are two sides of a coin; therefore, there is a strong relation between scale-frequency-saliency in visual attention problem. Several researchers =-=[3,30,27,33]-=- outlined that fixated regions have high spatial contrast or showed that high-frequency edges allow stronger discrimination between fixateed over non-fixated points. In brief, they all come up with on... |

334 | Saliency detection: A spectral residual approach
- Hou, Zhang
- 2007
(Show Context)
Citation Context ...ear contrast between features of central and surrounding environments across multiple scales in the center-surround operation. Saliency is also modeled as phase difference in Fourier Transform Domain =-=[20]-=-, or saliency at each location depends on statistical modeling of the local feature distribution [32]. Though many approaches are mentioned in long and rich literature of visual saliency, only a few a... |

308 | A multiscale random field model for Bayesian image segmentation.
- Bouman, Shapiro
- 1994
(Show Context)
Citation Context ... i. In order to estimate the MAP p(cji |dj), we need to employ Bayesian approach and capture dependencies between dyadic squares at different scales. Though many approximation techniques [8],[9],[24],=-=[5]-=- are derived for a practical MAP, the Hidden Markov Tree (HMT) by Choi [11] is proven to be a feasible solution. Choi introduces hidden label tree modeling instead of estimating joint probability of h... |

275 | Graph-based visual saliency - Harel, Koch, et al. - 2007 |

262 | Wavelet thresholding via a Bayesian approach.
- Abramovich, Sapatinas, et al.
- 1998
(Show Context)
Citation Context ...te refers to a large variance Gaussian and a "‘small"’ state represents small variance Gaussian distribution. This Gaussian mixture closely matches marginal statistics observed in natural images [28],=-=[1]-=-, [10]. With the HMT, persistence of large or small coefficients are captured across scales using Markov-1 chain. It models dependencies between hidden states across scale in a tree structure, paralle... |

206 |
Adaptive Bayesian wavelet shrinkage.
- Chipman, Kolaczyk, et al.
- 1997
(Show Context)
Citation Context ...fers to a large variance Gaussian and a "‘small"’ state represents small variance Gaussian distribution. This Gaussian mixture closely matches marginal statistics observed in natural images [28],[1], =-=[10]-=-. With the HMT, persistence of large or small coefficients are captured across scales using Markov-1 chain. It models dependencies between hidden states across scale in a tree structure, parallel to t... |

184 | Bayesian tree-structured image modeling using wavelet- based hidden Markov models”
- Romberg, Baraniuk
- 1999
(Show Context)
Citation Context ... processing. The HMT model parametersM can be learned through the iterative expectation and maximization (EM) algorithm with cost O(n) per iteration [13] or predefined for a particular image category =-=[31]-=-. After the parametersM are estimated by EM algorithm, we need to compute statistical characteristics of wavelet coefficients given the wavelet transform w̃ of a test image x̃ and a set of HMT paramet... |

151 |
Early processing of visual information
- Marr
(Show Context)
Citation Context ...formation. Robustness and efficiency of biological devices, the eyes and their control systems, visual paths in the brain have amazed scientists and engineers for centuries. From Neisser [26] to Marr =-=[25]-=-, researchers have put intesive effort in discovering attention principles and engineering artificial systems with equivalent capability. For last two decades, this research field is dominated by visu... |

146 | Visual correlates of fixation selection: Effects of scale and time
- Tatler, Baddeley, et al.
- 2005
(Show Context)
Citation Context .... In signal processing, scale-space and spectral space are two sides of a coin; therefore, there is a strong relation between scale-frequency-saliency in visual attention problem. Several researchers =-=[3,30,27,33]-=- outlined that fixated regions have high spatial contrast or showed that high-frequency edges allow stronger discrimination between fixateed over non-fixated points. In brief, they all come up with on... |

124 |
A feature-integration theory of attention, Cognitive Psychol
- Treisman, Gelade
- 1980
(Show Context)
Citation Context ..., this research field is dominated by visual saliency principles which proposes an existence of a saliency map for attention guidance. The idea is further promoted in Feature Integration Theory (FIT) =-=[34]-=- which elaborates computational principles of saliency map generation with center-surround operators and basic image features such as intensity, orientation and colors. Then, Itti et al. [21] implemen... |

108 | Multiscale image segmentation, using wavelet-domain hidden markov models.
- Choi, Baraniuk
- 2001
(Show Context)
Citation Context ...proach and capture dependencies between dyadic squares at different scales. Though many approximation techniques [8],[9],[24],[5] are derived for a practical MAP, the Hidden Markov Tree (HMT) by Choi =-=[11]-=- is proven to be a feasible solution. Choi introduces hidden label tree modeling instead of estimating joint probability of highdimensional dyadic squares. Due to strong correlation between the consid... |

97 | Natural scene statistics at the centre of gaze
- Reinagel, Zador
- 1999
(Show Context)
Citation Context .... In signal processing, scale-space and spectral space are two sides of a coin; therefore, there is a strong relation between scale-frequency-saliency in visual attention problem. Several researchers =-=[3,30,27,33]-=- outlined that fixated regions have high spatial contrast or showed that high-frequency edges allow stronger discrimination between fixateed over non-fixated points. In brief, they all come up with on... |

96 | On the plausibility of the discriminant center-surround hypothesis for visual saliency.
- Gao, Mahadevan, et al.
- 2008
(Show Context)
Citation Context ...tinguishing variances. It is reasonable to name "‘large"’ and "‘small"’ states according to their comparison in terms of variance values. Now the only difference between GSM models and Gao’s proposal =-=[14]-=- are whether GD or GGD should be used. Though GGD is more sophisticated with distribution shape parameter β, several factors support validity of simple GD modeling given the class labels as hidden var... |

72 | Multiresolution image classification by hierarchical modeling with two-dimensional hidden Markov models,”
- Li, Gray, et al.
- 2000
(Show Context)
Citation Context ...ation i. In order to estimate the MAP p(cji |dj), we need to employ Bayesian approach and capture dependencies between dyadic squares at different scales. Though many approximation techniques [8],[9],=-=[24]-=-,[5] are derived for a practical MAP, the Hidden Markov Tree (HMT) by Choi [11] is proven to be a feasible solution. Choi introduces hidden label tree modeling instead of estimating joint probability ... |

71 | Object-based visual attention for computer vision.
- Sun, Fisher
- 2003
(Show Context)
Citation Context ...s of visual attention with saliency maps, but does not investigate how such principles would be implemented algorithmically. It leaves research field open for many later saliency algorithms [21],[16],=-=[32]-=-,[19], etc. Saliency might be computed as a linear contrast between features of central and surrounding environments across multiple scales in the center-surround operation. Saliency is also modeled a... |

64 | Discriminant saliency for visual recognition from cluttered scenes.
- Gao, Vasconcelos
- 2005
(Show Context)
Citation Context ...spects of visual attention with saliency maps, but does not investigate how such principles would be implemented algorithmically. It leaves research field open for many later saliency algorithms [21],=-=[16]-=-,[32],[19], etc. Saliency might be computed as a linear contrast between features of central and surrounding environments across multiple scales in the center-surround operation. Saliency is also mode... |

42 | High frequency edges (but not contrast) predict where we fixate: A Bayesian system identification analysis.
- Baddeley, Tatler
- 2006
(Show Context)
Citation Context |

41 | Quantitative analysis of human-model agreement in visual saliency modeling: A comparative study.
- Borji, Sihite, et al.
- 2013
(Show Context)
Citation Context ...erical tools are necessary for simulation data analysis. In regards of fairness and accuracy of the evaluation, we employ a set of three measurements LCC, NSS, and AUC recommended by Ali Borji et al. =-=[4]-=- since evaluation codes can be retrieved freely from the authors’ website 6. Three evaluation scores are used for analyzing a method to ensure the reliability of qualitative conclusion and any conclus... |

34 | Bayesian approach to best basis selection
- Pesquet, Krim, et al.
- 1996
(Show Context)
Citation Context ...’ state refers to a large variance Gaussian and a "‘small"’ state represents small variance Gaussian distribution. This Gaussian mixture closely matches marginal statistics observed in natural images =-=[28]-=-,[1], [10]. With the HMT, persistence of large or small coefficients are captured across scales using Markov-1 chain. It models dependencies between hidden states across scale in a tree structure, par... |

27 | Decision-theoretic saliency: computational principle, biological plausibility, and implications for neurophysiology and psychophysics.
- Gao, Vasconcelos
- 2009
(Show Context)
Citation Context ...imated as saliency values for each location. This formulation of visual saliency approach is named as Discriminant Saliency (DIS) of which underlying principles are carefully elaborated by Gao et al. =-=[18]-=-. Its significant point is estimating information from class distributions given input features rather than from the input features themselves. Therefore, computational load is significantly reduced s... |

21 | Multiscale texture segmentation using wavelet-domain hidden Markov models
- Choi, Baraniuk
- 1999
(Show Context)
Citation Context .... A possible solution for modeling these relations between squares is using a general probabilistic graph [13]; however, the complexity exponentially increases with number of neighborhood nodes. Choi =-=[12]-=- proposes alternative simpler solution based on context-based Bayesian approach. For the sake of simplicity, causal contexts are only defined by states of the direct parent node and its 8 intermediate... |

17 |
Trainable context model for multiscale segmentation
- Cheng, Bouman
- 1998
(Show Context)
Citation Context ... location i. In order to estimate the MAP p(cji |dj), we need to employ Bayesian approach and capture dependencies between dyadic squares at different scales. Though many approximation techniques [8],=-=[9]-=-,[24],[5] are derived for a practical MAP, the Hidden Markov Tree (HMT) by Choi [11] is proven to be a feasible solution. Choi introduces hidden label tree modeling instead of estimating joint probabi... |

14 |
Saliency based on information maximization. Advances in neural information processing systems
- Bruce, Tsotsos
- 2006
(Show Context)
Citation Context ... are mentioned in long and rich literature of visual saliency, only a few are built on a solid theory or linked to other well-established computational theory. Among the approaches, Neil Bruce’s work =-=[6]-=- nicely established a bridge between visual saliency and information theory. It put a first step for bridging two alien fields; moreover, visual attention for first time could be modeled as informatio... |

10 |
Simplified wavelet-domain hidden Markov models using contexts
- Crouse, Baraniuk
- 1997
(Show Context)
Citation Context ...r modeling natural images. In fact, statistical modeling of wavelet coefficients have quite comprehensive literature; here we concentrate on the hidden Markov tree (HMT) of Crouse, Nowak and Baraniuk =-=[13]-=-. In consideration of both marginal and joint wavelet coefficient statistics, the HMT model introduces a (hidden) state variable of either "‘large"’ or "‘small"’ to each wavelet coefficients. Then, th... |

5 | Tsotsos, Visual correlates of fixation selection: a look at the spatial frequency domain
- Bruce, Loach, et al.
(Show Context)
Citation Context ...quencies. Though these studies emphasizes a greater visual attraction to high frequencies (edges, ridges, other structures of images), there are other works focusing on medium frequency. Bruce et al. =-=[7]-=- found that fixation points tend to prefere horizontal and vertical frequency content rather than random position, and these oriented content have more noticeable difference in medium frequencies. Mor... |

5 |
Discriminant interest points are stable
- Gao, Vasconcelos
(Show Context)
Citation Context ...stribution (GGD) and only binary classification is considered. Lets name Gao’s proposal for discriminant saliency computation as DIS; more details about DIS can be found in their publications [18][14]=-=[17]-=-[15]. While DIS successfully defines discriminant saliency in information-theoretic senses, its implementation has certain limits. Feature responses are randomly sampled in a single fixed-size window;... |

5 | An information theoretic model of spatiotemporal visual saliency
- Qiu, Gu, et al.
(Show Context)
Citation Context ... moreover, visual attention for first time could be modeled as information system. Then, information-based visual saliency has continuously been investigated and developed in several works [23],[22], =-=[29]-=-, [14]. The distinguish points between these works are computational approaches for information retrieval from features. The process attracts much interest due to difficulty in estimating information ... |

3 |
Visual saliency based on fast nonparametric multidimensional entropy estimation
- Ngo, Qiu, et al.
(Show Context)
Citation Context ...ien fields; moreover, visual attention for first time could be modeled as information system. Then, information-based visual saliency has continuously been investigated and developed in several works =-=[23]-=-,[22], [29], [14]. The distinguish points between these works are computational approaches for information retrieval from features. The process attracts much interest due to difficulty in estimating i... |

2 |
Effects of luminance contrast and its modifications on fixation behavior during free viewing of images from different categories, Vision research 49 (12) (2009) 1541. URL http://www.sonat.bonobostree.com/doc/pdf/acik_vr_2009. pdf
- A\ccık, Onat, et al.
(Show Context)
Citation Context ...ition, and these oriented content have more noticeable difference in medium frequencies. More interestingly, choices of frequency range for visual processing may depend on encountering visual context =-=[2]-=-. For example, luminance contrast explained fixation locations better in natural image category and slightly worse in urban scenes category provided that all images are applied low-pass filters as pre... |

2 | Multiscale document segmentation 1
- Cheng, Bouman, et al.
- 1997
(Show Context)
Citation Context ... and location i. In order to estimate the MAP p(cji |dj), we need to employ Bayesian approach and capture dependencies between dyadic squares at different scales. Though many approximation techniques =-=[8]-=-,[9],[24],[5] are derived for a practical MAP, the Hidden Markov Tree (HMT) by Choi [11] is proven to be a feasible solution. Choi introduces hidden label tree modeling instead of estimating joint pro... |

2 |
Improvement and evaluation of visual saliency based on information theory
- Ngo, Ang, et al.
(Show Context)
Citation Context ...ields; moreover, visual attention for first time could be modeled as information system. Then, information-based visual saliency has continuously been investigated and developed in several works [23],=-=[22]-=-, [29], [14]. The distinguish points between these works are computational approaches for information retrieval from features. The process attracts much interest due to difficulty in estimating inform... |