Information Complexity Ranking: A New Method of Ranking Images by Algorithmic Complexity.

Area of Science:

• Computer Science
• Information Theory
• Human-Computer Interaction

Background:

• Predicting visual complexity is crucial for user interface design but remains underexplored.
• Existing methods lack a comprehensive approach to intrinsic and relational complexity.

Purpose of the Study:

• To introduce a novel method, Information Complexity Ranking (ICR), for ranking visual information complexity.
• To evaluate ICR's performance against established algorithms using diverse image datasets.

Main Methods:

• Developed ICR, integrating Kolmogorov complexity (intrinsic algorithmic complexity) and normalized compression distance (NCD) for object similarity.
• Validated ICR on 7200 randomly generated images (text, digits, colored dots).
• Tested ICR on 1400 categorized images, comparing results with five state-of-the-art complexity algorithms.

Main Results:

• ICR demonstrated superior performance in certain image categories compared to existing algorithms.
• The method showed competitive results across various categories, outperforming the majority of state-of-the-art approaches.
• ICR's efficiency decreased for images with high semantic content.

Conclusions:

• Information Complexity Ranking (ICR) offers a promising approach to quantifying visual complexity.
• The method effectively balances intrinsic and relational complexity measures.
• Further refinement is needed for complex, semantically rich visual information.