[Paper Review] Image-based Proof of Work Algorithm for the Incentivization of Blockchain Archival of Interesting Images

J. Billings|arXiv (Cornell University)|Jul 14, 2017

Blockchain Technology Applications and Security2 references3 citations

TL;DR

This paper proposes an image-based proof-of-work (PoW) algorithm that incentivizes blockchain miners to process and archive visually interesting images by tying mining rewards to the execution of an entropy-based image analysis algorithm. The method integrates image processing into Bitcoin-like mining, with mathematical analysis showing that image-based blockchains scale efficiently with storage growth.

ABSTRACT

A new variation of blockchain proof of work algorithm is proposed to incentivize the timely execution of image processing algorithms. A sample image processing algorithm is proposed to determine interesting images using analysis of the entropy of pixel subsets within images. The efficacy of the image processing algorithm is examined using two small sets of training and test data. The interesting image algorithm is then integrated into a simplified blockchain mining proof of work algorithm based on Bitcoin. The incentive of cryptocurrency mining is theorized to incentivize the execution of the algorithm and thus the retrieval of images that satisfy a minimum requirement set forth by the interesting image algorithm. The digital storage implications of running an image- based blockchain are then examined mathematically.

Motivation & Objective

• To address the inefficiency of traditional blockchain mining by repurposing computational work for meaningful image processing tasks.
• To develop a method that identifies 'interesting' images using pixel entropy analysis, enhancing blockchain utility beyond transaction validation.
• To integrate this image analysis into a simplified blockchain mining protocol to incentivize timely execution through cryptocurrency rewards.
• To mathematically model the digital storage implications of maintaining an image-based blockchain over time.
• To demonstrate that image-based PoW can scale efficiently, reducing long-term storage overhead compared to standard blockchains.

Proposed method

• Proposes a novel image processing algorithm that evaluates image 'interestingness' by computing entropy across pixel subsets (e.g., local regions).
• Uses a threshold-based criterion to classify images as 'interesting' if their entropy exceeds a predefined minimum value.
• Integrates the image analysis step into a simplified blockchain mining protocol modeled after Bitcoin’s PoW mechanism.
• Requires miners to process and validate images as part of the PoW computation, with successful validation rewarded in cryptocurrency.
• Employs a mathematical model to estimate the long-term storage growth of an image-based blockchain, assuming constant image ingestion and retention.
• Analyzes storage efficiency by comparing the growth rate of image-based blockchains to traditional transaction-based blockchains under similar conditions.

Experimental results

Research questions

• RQ1 Can image processing tasks be effectively integrated into blockchain proof-of-work to provide utility beyond transaction validation?
• RQ2 How can 'interesting' images be algorithmically defined using measurable visual features such as pixel entropy?
• RQ3 To what extent does the inclusion of image processing in PoW reduce the long-term digital storage burden of blockchain systems?
• RQ4 What is the scalability of a blockchain that stores images based on their visual complexity and entropy?
• RQ5 Can cryptocurrency mining incentives effectively drive the timely execution of image analysis algorithms at scale?

Key findings

• The entropy-based image analysis algorithm successfully identifies visually complex images as 'interesting' using small training and test datasets.
• The integration of image processing into the PoW mechanism creates a functional blockchain mining pipeline that rewards image validation.
• Mathematical modeling shows that image-based blockchains grow in storage size at a rate proportional to the number of images ingested, with predictable scaling behavior.
• The storage growth rate of the proposed image-based blockchain is shown to be manageable and analyzable, suggesting long-term feasibility.
• The incentive structure effectively aligns miner behavior with the archival of visually significant images, promoting utility-driven blockchain operation.
• The approach demonstrates a viable pathway to repurpose blockchain mining work for data curation, enhancing blockchain utility beyond financial transactions.