Container File Grouping for Semantic Image Tracks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The MPEG-H Image File Format lacks a consistent mechanism for grouping semantically related image items and image sequence tracks, making it difficult to indicate relationships between images and tracks, such as burst images and high-quality images, or multi-view images, which affects how contents are presented.
Innovation Solution
A new container file format mechanism is introduced that allows grouping of static media items and timed media tracks, using a 'GroupsList' box and 'EntityToGroup' boxes to specify grouping types and relationships between items and tracks, enabling alternate, preview, origin, multiview, and equivalent groupings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If no grouping mechanism is provided in the MPEG-H Image File Format, then the file structure remains simple and straightforward, but it becomes impossible to indicate semantic relationships between image items and image sequence tracks
Solution Approach 1:
The patent introduces a hierarchical grouping structure that segments the file format into distinct components: GroupsList box at the top level, EntityToGroup boxes for individual entity-group mappings, and nested subgroups. This segmentation allows semantic relationship information to be organized in a structured manner without overwhelming the overall file format complexity.
Solution Approach 2:
The patent introduces intermediate data structures (GroupsList box, EntityToGroup boxes, and grouping type indicators) that act as mediators between image items and image sequence tracks. These intermediaries enable the indication of semantic relationships without directly modifying the core image data structures, thus maintaining simplicity while adding functionality.
2Adaptability or versatility
If multiple separate mechanisms are used for different grouping scenarios (burst images, multi-view, preview), then each scenario can be handled specifically, but the file format becomes more complex and harder to process
Solution Approach 1:
The patent creates a universal grouping mechanism using EntityToGroup boxes that can handle multiple grouping scenarios (burst images, multi-view, preview, alternate representations) through a single unified data structure. The versatility is achieved by varying the grouping type indicator and entity identifiers rather than creating separate mechanisms for each scenario, thereby reducing processing complexity while maintaining adaptability.
3Loss of information
If a detailed grouping mechanism is introduced to handle all semantic relationships, then relationship indication becomes possible, but file size and processing overhead increase
Solution Approach 1:
The patent uses efficient data representation by copying only essential identifiers (entity identifiers and grouping type indicators) rather than duplicating full image data or detailed relationship descriptions. The EntityToGroup boxes contain compact references to entities and their group memberships, minimizing the additional data volume required to encode semantic relationships.
Solution Approach 2:
The patent encodes semantic relationships by changing parameters within the existing file structure (adding grouping type fields and entity identifiers) rather than fundamentally altering the data format. This approach adds minimal overhead by utilizing efficient parameter encoding schemes that require small amounts of additional data to represent complex relationships.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
There are disclosed various methods, apparatuses and computer program products for video encoding. According to an embodiment,an encoding method comprises including a static media item into a container file; including one or more media tracks into the container file; indicating, in the file, that the static media item and one or more entities form a group; and indicating, in the file, a grouping type for the group. According to an embodiment, a decoding method comprises parsing, from a container file, that a static media item and one or more entities form a group; parsing, from the container file, a grouping type for the group; determining a processing for the static media item and the one or more of the entities based on the group and the grouping type.