Compressible Earth Mover's Distance for Image Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image matching technologies, such as Earth Mover's Distance (EMD), do not accurately reflect human perception of distance between image distributions, leading to inefficiencies in image comparison and retrieval.
Innovation Solution
The Compressible Earth Mover's Distance (CEMD) method, which allows for the compression of 'earth' and 'holes' when moving them, thereby minimizing distance and improving image comparison accuracy by incorporating a penalty for compression, is used to calculate the work required to transform one distribution into another.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Earth Mover's Distance (EMD) is used to measure image similarity, then image comparison can be performed, but the measurement does not accurately reflect human perception of distance between image distributions
Solution Approach 1:
The patent modifies the EMD calculation by introducing compressibility parameters that allow earth and holes to be compressed when moving between locations. This changes the mathematical parameters of the distance calculation to better reflect human perception of image similarity, where small local variations are less significant than overall distribution patterns.
2Measurement precision
If compression of earth and holes is allowed in CEMD calculation, then the distance measurement better matches human perception, but the computational complexity increases
Solution Approach 1:
The patent applies compression selectively rather than uniformly across all earth and hole movements. By allowing compression only when it improves the match between earth and hole distributions, the method achieves better human perception alignment without requiring exhaustive computation of all possible compression scenarios.
Solution Approach 2:
The patent pre-calculates compression factors and stores them for use during the CEMD computation. This preliminary preparation reduces the computational burden during the actual image comparison by avoiding repeated calculation of compression parameters.
Data Source
AI summary
A Compressible Earth Mover's Distance (CEMD) better matches how humans perceive distance between distributions. Earth and holes are able to be compressed when moving from the earth to the holes thus minimizing the distance and improving the quality of image comparison. CEMD is utilized in a number of implementations, for instance, content based image retrieval, color query and other applications in multimedia. Another implementation includes using CEMD with a content recognition system for indexing occurrences of objects within an audio/video content data stream which processes the stream of data to generate a content index database corresponding to the content stream. CEMD is usable with in a variety of systems to assist in image recognition.