Visual attention model based on statistical properties of neuron responses.

Area of Science:

• Neuroscience
• Computer Vision
• Computational Neuroscience

Background:

• Visual attention selects relevant objects within a scene.
• Neural interactions in cortical areas are crucial for attentional allocation.
• Neuron response characteristics in attention-related cortices remain unclear.

Purpose of the Study:

• To demonstrate that unusual regions attracting more attention elicit specific neuron responses.
• To propose a bottom-up visual attention model based on neuron responses to natural scene contexts.
• To investigate the neural mechanisms underlying early visual cortex function in bottom-up attention.

Main Methods:

• Developed a bottom-up visual attention model utilizing self-information of neuron responses.
• Employed four distinct color spaces and a novel entropy-based scheme for color information integration.
• Generated saliency maps highlighting important regions and suppressing backgrounds.

Main Results:

• The proposed model effectively highlights salient regions in natural scenes.
• Comparative analysis showed the model's superiority over several state-of-the-art models.
• Saliency maps demonstrated effective suppression of redundant background information.

Conclusions:

• Visual saliency is likely derived from neuron responses to contextual information in natural scenes.
• The proposed model offers insights into neuron response-based saliency detection.
• Findings may elucidate the neural mechanisms of bottom-up visual attention in early visual cortices.