Exploiting Surroundedness for Saliency Detection: A Boolean Map Approach.

Area of Science:

• Computer Vision
• Computational Neuroscience
• Human-Computer Interaction

Background:

• Eye fixation prediction is crucial for understanding visual attention.
• Existing models often struggle to capture complex visual cues effectively.

Purpose of the Study:

• To introduce a novel saliency model, Boolean Map based Saliency (BMS), for enhanced eye fixation prediction.
• To demonstrate the effectiveness of the 'surroundedness' cue in saliency mapping.

Main Methods:

• Image representation using binary maps derived from feature maps.
• Saliency map computation via topological analysis of Boolean maps to identify surrounded regions.
• Connection established between BMS and Minimum Barrier Distance for theoretical insight.

Main Results:

• The BMS model effectively utilizes the surroundedness cue.
• BMS demonstrates superior performance compared to 10 state-of-the-art methods.
• The model is validated on seven diverse eye-tracking benchmark datasets.

Conclusions:

• BMS offers a simple, efficient, and high-performing approach to eye fixation prediction.
• The 'surroundedness' cue, captured through Boolean maps, is a powerful predictor of visual attention.
• The proposed method advances the field of computational saliency and visual attention modeling.