Image Metadata Creation Using Pyramidal Storage Structures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for creating metadata from image and video data often require manipulation of the source data, altering its integrity, and lack efficient storage and usage strategies for robust functionality.
Innovation Solution
A media processor that computes and stores metadata from image and video data using statistical attributes, generating multi-scale versions and storing them in efficient data structures like pyramidal structures, without altering the source data, enabling various applications such as display management, DRM, and content verification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If metadata is created by manipulating source image data, then processing functionality is improved, but source data integrity is compromised
Solution Approach 1:
The patent separates metadata creation from source data manipulation by computing statistical attributes (histograms, moments, etc.) as independent derived data structures. This segmentation allows metadata to be used for various processing functions while the source image data remains untouched and intact.
2Measurement precision
If full-resolution metadata is stored for all image data, then processing accuracy is improved, but storage efficiency deteriorates
Solution Approach 1:
The patent applies local quality by creating multi-scale metadata representations where different levels of detail are stored for different regions or resolutions. Full-resolution metadata is only stored where needed, while lower-resolution metadata suffices for other areas, optimizing the balance between processing accuracy and storage efficiency.
Solution Approach 2:
The system computes and stores metadata at multiple scales, providing partial metadata at lower resolutions and full metadata only at higher resolutions when needed. This partial action approach reduces overall storage requirements while maintaining processing accuracy where necessary.
3Adaptability or versatility
If comprehensive image statistics are collected for all possible processing applications, then system versatility is improved, but processing complexity increases
Solution Approach 1:
The patent creates a universal metadata framework where a single set of statistical attributes (histograms, moment invariants, etc.) can serve multiple processing applications including display management, DRM verification, and content analysis. This multi-functional metadata structure reduces the need for separate processing pipelines for different applications.
Solution Approach 2:
The system pre-computes comprehensive image statistics and stores them as metadata before actual processing operations are performed. This preliminary action allows various processing applications to utilize the pre-prepared metadata without performing redundant computations, thereby reducing processing complexity during actual operations.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
Several embodiments of a media processor for the creation of metadata from a set of images are given. The media processor receives a set of image data and computes metadata from the image data. The metadata may comprise statistical data based on the image data; from which the media processor may generate and store a multi-scale version of the metadata. The metadata may comprise statistical attributes derived directly from the image data or statistical attributes derived from image appearance correlates of the image data. The media processor may generate a subsampled set of the metadata and the subsampled set may be stored in an efficient data structure, such as a pyramidal structure. The pyramidal structure may be generated from spatial image information, from temporal image information or from spatial-temporal image information.