Limits...
Visual attention model based on statistical properties of neuron responses.

Duan H, Wang X - Sci Rep (2015)

Bottom Line: Valuable regions are highlighted while redundant backgrounds are suppressed in the saliency maps obtained by the proposed model.Comparative results reveal that the proposed model outperforms several state-of-the-art models.This study provides insights into the neuron responses based saliency detection and may underlie the neural mechanism of early visual cortices for bottom-up visual attention.

View Article: PubMed Central - PubMed

Affiliation: 1] State Key Laboratory of Virtual Reality Technology and Systems, Beihang University, Beijing 100191, P. R. China [2] Science and Technology on Aircraft Control Laboratory, School of Automation Science and Electronic Engineering, Beihang University, Beijing 100191, P. R. China.

ABSTRACT
Visual attention is a mechanism of the visual system that can select relevant objects from a specific scene. Interactions among neurons in multiple cortical areas are considered to be involved in attentional allocation. However, the characteristics of the encoded features and neuron responses in those attention related cortices are indefinite. Therefore, further investigations carried out in this study aim at demonstrating that unusual regions arousing more attention generally cause particular neuron responses. We suppose that visual saliency is obtained on the basis of neuron responses to contexts in natural scenes. A bottom-up visual attention model is proposed based on the self-information of neuron responses to test and verify the hypothesis. Four different color spaces are adopted and a novel entropy-based combination scheme is designed to make full use of color information. Valuable regions are highlighted while redundant backgrounds are suppressed in the saliency maps obtained by the proposed model. Comparative results reveal that the proposed model outperforms several state-of-the-art models. This study provides insights into the neuron responses based saliency detection and may underlie the neural mechanism of early visual cortices for bottom-up visual attention.

Show MeSH

Related in: MedlinePlus

Framework of the proposed visual attention model.Each rectangle depicts an operation involved in the overall computational framework. Color space transformation: Original image saved in the RGB color space is transformed into gray-scale image. It is also transformed into the CIELAB, HSI, and YIQ color spaces. Channels of all the four color spaces are separated. Compute neuron responses: Neurons responses in area V1 are extracted from local patches of gray-scale image and all the twelve separate channels of the four color spaces (see Methods). Afterwards, statistical probabilities of the neuron responses are estimated based on the histogram of the neuron responses. Compute visual saliency: Saliency sub-maps are obtained by adding the self-information in each dimension of the neuron responses up. Saliency sub-maps are computed for gray-scale image and all the twelve separate channels. Combine into final saliency map: The entropy of each saliency sub-map is computed. Afterwards the sub-map with the lowest entropy is selected from each color space (the sub-maps in red rectangles). Finally, the selected sub-maps of all the four color spaces together with the sub-map of the gray-scale image are combined into the final saliency map. Reciprocals of their corresponding entropies are taken as combining weights for sub-maps. The original image is taken by X.H. Wang with a digital camera Canon IXUS 125HS.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4352866&req=5

f1: Framework of the proposed visual attention model.Each rectangle depicts an operation involved in the overall computational framework. Color space transformation: Original image saved in the RGB color space is transformed into gray-scale image. It is also transformed into the CIELAB, HSI, and YIQ color spaces. Channels of all the four color spaces are separated. Compute neuron responses: Neurons responses in area V1 are extracted from local patches of gray-scale image and all the twelve separate channels of the four color spaces (see Methods). Afterwards, statistical probabilities of the neuron responses are estimated based on the histogram of the neuron responses. Compute visual saliency: Saliency sub-maps are obtained by adding the self-information in each dimension of the neuron responses up. Saliency sub-maps are computed for gray-scale image and all the twelve separate channels. Combine into final saliency map: The entropy of each saliency sub-map is computed. Afterwards the sub-map with the lowest entropy is selected from each color space (the sub-maps in red rectangles). Finally, the selected sub-maps of all the four color spaces together with the sub-map of the gray-scale image are combined into the final saliency map. Reciprocals of their corresponding entropies are taken as combining weights for sub-maps. The original image is taken by X.H. Wang with a digital camera Canon IXUS 125HS.

Mentions: As Figure 1 suggests, appearances in different color spaces reveal distinct information of an image. Four color spaces are used in the proposed model to take advantages of color information. The saliency sub-map in each channel is computed on the foundation of corresponding neuron responses. The final saliency map is obtained via a novel entropy-based combination operation. The red car in Figure 1 is highlighted while the plants and road in the background are suppressed in the final saliency map, revealing the effect of visual attention.


Visual attention model based on statistical properties of neuron responses.

Duan H, Wang X - Sci Rep (2015)

Framework of the proposed visual attention model.Each rectangle depicts an operation involved in the overall computational framework. Color space transformation: Original image saved in the RGB color space is transformed into gray-scale image. It is also transformed into the CIELAB, HSI, and YIQ color spaces. Channels of all the four color spaces are separated. Compute neuron responses: Neurons responses in area V1 are extracted from local patches of gray-scale image and all the twelve separate channels of the four color spaces (see Methods). Afterwards, statistical probabilities of the neuron responses are estimated based on the histogram of the neuron responses. Compute visual saliency: Saliency sub-maps are obtained by adding the self-information in each dimension of the neuron responses up. Saliency sub-maps are computed for gray-scale image and all the twelve separate channels. Combine into final saliency map: The entropy of each saliency sub-map is computed. Afterwards the sub-map with the lowest entropy is selected from each color space (the sub-maps in red rectangles). Finally, the selected sub-maps of all the four color spaces together with the sub-map of the gray-scale image are combined into the final saliency map. Reciprocals of their corresponding entropies are taken as combining weights for sub-maps. The original image is taken by X.H. Wang with a digital camera Canon IXUS 125HS.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4352866&req=5

f1: Framework of the proposed visual attention model.Each rectangle depicts an operation involved in the overall computational framework. Color space transformation: Original image saved in the RGB color space is transformed into gray-scale image. It is also transformed into the CIELAB, HSI, and YIQ color spaces. Channels of all the four color spaces are separated. Compute neuron responses: Neurons responses in area V1 are extracted from local patches of gray-scale image and all the twelve separate channels of the four color spaces (see Methods). Afterwards, statistical probabilities of the neuron responses are estimated based on the histogram of the neuron responses. Compute visual saliency: Saliency sub-maps are obtained by adding the self-information in each dimension of the neuron responses up. Saliency sub-maps are computed for gray-scale image and all the twelve separate channels. Combine into final saliency map: The entropy of each saliency sub-map is computed. Afterwards the sub-map with the lowest entropy is selected from each color space (the sub-maps in red rectangles). Finally, the selected sub-maps of all the four color spaces together with the sub-map of the gray-scale image are combined into the final saliency map. Reciprocals of their corresponding entropies are taken as combining weights for sub-maps. The original image is taken by X.H. Wang with a digital camera Canon IXUS 125HS.
Mentions: As Figure 1 suggests, appearances in different color spaces reveal distinct information of an image. Four color spaces are used in the proposed model to take advantages of color information. The saliency sub-map in each channel is computed on the foundation of corresponding neuron responses. The final saliency map is obtained via a novel entropy-based combination operation. The red car in Figure 1 is highlighted while the plants and road in the background are suppressed in the final saliency map, revealing the effect of visual attention.

Bottom Line: Valuable regions are highlighted while redundant backgrounds are suppressed in the saliency maps obtained by the proposed model.Comparative results reveal that the proposed model outperforms several state-of-the-art models.This study provides insights into the neuron responses based saliency detection and may underlie the neural mechanism of early visual cortices for bottom-up visual attention.

View Article: PubMed Central - PubMed

Affiliation: 1] State Key Laboratory of Virtual Reality Technology and Systems, Beihang University, Beijing 100191, P. R. China [2] Science and Technology on Aircraft Control Laboratory, School of Automation Science and Electronic Engineering, Beihang University, Beijing 100191, P. R. China.

ABSTRACT
Visual attention is a mechanism of the visual system that can select relevant objects from a specific scene. Interactions among neurons in multiple cortical areas are considered to be involved in attentional allocation. However, the characteristics of the encoded features and neuron responses in those attention related cortices are indefinite. Therefore, further investigations carried out in this study aim at demonstrating that unusual regions arousing more attention generally cause particular neuron responses. We suppose that visual saliency is obtained on the basis of neuron responses to contexts in natural scenes. A bottom-up visual attention model is proposed based on the self-information of neuron responses to test and verify the hypothesis. Four different color spaces are adopted and a novel entropy-based combination scheme is designed to make full use of color information. Valuable regions are highlighted while redundant backgrounds are suppressed in the saliency maps obtained by the proposed model. Comparative results reveal that the proposed model outperforms several state-of-the-art models. This study provides insights into the neuron responses based saliency detection and may underlie the neural mechanism of early visual cortices for bottom-up visual attention.

Show MeSH
Related in: MedlinePlus