Image Data Encapsulation With Tile Metadata for Spatial Access
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image file formats, such as the ISO/IEC 23008-12 standard, lack sufficient descriptive metadata for efficiently accessing and extracting tiles from high-resolution images and video data, particularly for applications like computational photography and video summarization, and do not support non-regular grids with existing EXIF tags.
Innovation Solution
The method involves providing tile description information with spatial parameters for dividing images into tiles, associating tile picture items with reference information, and outputting the encoded bitstream as an encapsulated data file, extending the metadata boxes to support tile-based access and region of interest extraction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If existing image file formats (ISO/IEC 23008-12) are used for storing high-resolution images, then the images can be stored with compressed data, but the lack of descriptive metadata makes it difficult to efficiently access and extract tiles or regions of interest
Solution Approach 1:
The patent divides the image into multiple tiles using spatial parameters (tile_left, tile_top, tile_width, tile_height) that define the position and size of each tile. This segmentation allows the image to be accessed in manageable portions, enabling efficient extraction of regions of interest without processing the entire high-resolution image at once.
Solution Approach 2:
The patent pre-processes the image by embedding descriptive metadata boxes (such as 'apic' for spatial parameters, 'stts' for time information, and custom tile description boxes) into the compressed bitstream during encoding. This preliminary organization of information allows for rapid random access and extraction of specific tiles without requiring complex processing during playback.
2Productivity
If metadata boxes are extended to support tile-based access, then efficient extraction of tiles becomes possible, but the device complexity increases
Solution Approach 1:
The patent extends existing metadata box structures (such as the 'apic' box for spatial parameters and 'stts' box for time information) to serve multiple functions: describing both the complete image and individual tiles within the image. This multi-functionality allows the same metadata infrastructure to support both full-image access and tile-level extraction, reducing the need for entirely separate complex systems.
Solution Approach 2:
The patent embeds tile description information within the existing metadata box hierarchy, creating a nested structure where tile-level metadata is contained within or referenced by higher-level image metadata boxes. This nesting approach allows the metadata to be organized in a hierarchical manner, where tile information is integrated into the overall image description without requiring completely separate metadata structures.
3Ease of operation
If tile description information is embedded in the bitstream, then easy identification and extraction of tiles is achieved, but the bitstream size increases
Solution Approach 1:
The patent uses compact parameter representations for tile descriptions, encoding spatial parameters (tile_left, tile顶, tileWidth, tileHeight) and temporal parameters (firstTime, secondTime) in an efficient format. By changing the parameter encoding to use minimal necessary bits while maintaining precision, the patent reduces the overhead added to the bitstream while still enabling accurate tile identification and extraction.
Data Source
AI summary
There is provided a method of encapsulating an encoded bitstream representing one or more images, the encapsulated bitstream comprising a data part and a metadata part. The method comprises:providing image item information identifying a portion of the data part representing a sub-image or an image of a single image and/or a set of single images;providing image description information comprising parameters including display parameters and/or transformation operators relating to one or more images andoutputting said bitstream together with said provided information as an encapsulated data file.Said image item information comprises one or more properties including at least part of the image description information dedicated to the considered sub-image or single image or set of single images, said image description information being defined in one or more boxes.