What Makes an Image Interesting and How Can We Explain It

Area of Science:

• Cognitive Science
• Computer Vision
• Psychology

Background:

• Understanding what makes a photograph interesting is key to visual cognition.
• This knowledge can improve information dissemination.
• Previous research has not fully explained image interestingness.

Purpose of the Study:

• To investigate the intrinsic properties that determine a photograph's interestingness.
• To identify factors influencing human perception of image interest.
• To explore the relationship between image characteristics and observer ratings.

Main Methods:

• Observers rated the interestingness of diverse real-world scene images.
• Image categories included landscapes, people, cityscapes, aerial, and indoor scenes.
• Statistical models analyzed image properties and eye-tracking data.

Main Results:

• A clear hierarchy of interestingness was found: landscapes > people/cityscapes > aerial > indoor scenes.
• Image interestingness is an intrinsic property, unaffected by viewing duration or history.
• Image uprightness significantly predicted interestingness; low-level features like complexity and color did not.
• Eye movement patterns, including novel fixations, partially predicted interest levels.

Conclusions:

• A category-based hierarchy of image interestingness exists and is an intrinsic property.
• Image uprightness is a significant predictor of interestingness.
• Interestingness is distinct from memorability or awe and not fully explained by current appraisal theories.