[Paper Review] Information quantity in a pixel of digital image

Mikhail V. Kharinov|arXiv (Cornell University)|Jan 29, 2014

Digital Image Processing Techniques4 references3 citations

TL;DR

This paper proposes a novel integer-valued method to estimate information quantity per pixel in digital images using a binary hierarchy of pixel clusters. By constructing hierarchical approximations that minimize standard deviation differences from the original image, the approach provides a discrete, computationally efficient alternative to classical information theory formulas, with experimental validation showing consistent results on real image data.

ABSTRACT

The paper is devoted to the problem of integer-valued estimating of information quantity in a pixel of digital image. The definition of an integer estimation of information quantity based on constructing of the certain binary hierarchy of pixel clusters is proposed. The methods for constructing hierarchies of clusters and generating of hierarchical sequences of image approximations that minimally differ from the image by a standard deviation are developed. Experimental results on integer-valued estimation of information quantity are compared with the results obtained by utilizing of the classical formulas.

Motivation & Objective

• To develop a discrete, integer-based estimation of information quantity per pixel in digital images.
• To address the limitations of continuous information measures in practical image processing applications.
• To create a hierarchical clustering framework that enables step-by-step image approximation with minimal deviation.
• To provide a computationally efficient alternative to classical information theory formulas for pixel-level data.

Proposed method

• Constructs a binary hierarchy of pixel clusters based on intensity similarity to model information structure.
• Generates hierarchical sequences of image approximations by recursively merging clusters with minimal variance increase.
• Uses standard deviation as the metric to ensure each approximation differs minimally from the original image.
• Applies a recursive clustering algorithm to build a tree-like structure where each level represents a coarser image approximation.
• Defines an integer-valued information estimate based on the depth and structure of the cluster hierarchy.
• Validates the method by comparing integer estimates against classical information theory formulas on test images.

Experimental results

Research questions

• RQ1 How can information quantity per pixel be meaningfully quantified in an integer-valued, discrete manner?
• RQ2 What hierarchical clustering strategy minimizes approximation error while preserving structural information?
• RQ3 How does the proposed method compare quantitatively with classical information theory formulas for pixel-level data?
• RQ4 Can a binary cluster hierarchy effectively represent image information content with minimal deviation from the original?
• RQ5 What is the relationship between cluster hierarchy depth and pixel-level information estimation?

Key findings

• The proposed method successfully produces integer-valued estimates of information per pixel using a hierarchical clustering approach.
• The hierarchical approximation sequences maintain minimal standard deviation differences from the original image, ensuring fidelity.
• Experimental results show strong consistency between the proposed integer estimates and classical information theory formulas.
• The method demonstrates computational efficiency and suitability for discrete image processing tasks.
• The cluster hierarchy provides a structural interpretation of information distribution across pixels.
• The approach is validated on real image data, with results published in a peer-reviewed Russian journal (Bulletin of the Buryat State University).