Selective visual attention based clutter metric with human visual system adaptability

Area of Science:

• Computer Vision
• Human-Computer Interaction
• Visual Perception

Background:

• Existing clutter metrics rely on fixed features and subjective weights.
• Clutter significantly impacts target acquisition performance and visual search efficiency.

Purpose of the Study:

• To propose a novel clutter metric that accounts for selective visual attention effects.
• To develop a metric consistent with the adaptive nature of the human visual system (HVS).

Main Methods:

• Extracting adaptive structural features based on edge-structure similarity to the target.
• Selecting confusing blocks using an attention guidance map and similarity threshold.
• Quantifying clutter by measuring the impact of confusing blocks on target acquisition performance.

Main Results:

• The proposed clutter metric demonstrates consistency with HVS adaptability.
• Comparative experiments on the Search_2 dataset show superior performance over existing metrics.
• The metric effectively quantifies clutter based on attentional effects.

Conclusions:

• The novel clutter metric offers a more accurate assessment of visual clutter.
• This approach enhances understanding of clutter's impact on visual attention and search.
• The metric has potential applications in optimizing visual displays and search interfaces.