Image Processing Device Marker Thinning for Efficient Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In image processing, the large number of restart markers inserted with short intervals in data streams leads to an enormous amount of data when their addresses are maintained as a table, making efficient decoding processing challenging, especially for large images or when only a specific region of an image needs to be decoded.
Innovation Solution
An image processing device and method that thin out restart markers at predetermined intervals, aligning them vertically, and generate marker information with position details, allowing for efficient decoding by specifying only the necessary markers for the decoding target region, reducing the data required for decoding processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If restart markers are inserted with short intervals to enable segmentation of encoding processing, then decoding processing can be performed in the middle of the data stream, but the amount of data in the marker address table becomes enormous
Solution Approach 1:
The patent extracts only the necessary marker information needed for decoding operations. Instead of maintaining a complete table of all restart marker addresses, the system extracts and stores only the essential marker data required to perform decoding in the middle of the data stream, thereby reducing the overall data amount while preserving functionality.
Solution Approach 2:
The patent segments the marker address table into multiple pages, with each page containing a subset of marker addresses. This segmentation allows the system to load only the necessary pages into memory based on the decoding requirements, rather than loading the entire enormous table, thus reducing the active data amount in memory.
2Ease of manufacture
If all marker addresses are maintained in a table for direct access, then any restart marker can be accessed directly, but the table data amount becomes enormous for large images
Solution Approach 1:
The patent introduces a hierarchical structure with multiple dimensions for marker address storage. Instead of a single flat table, the system uses a multi-page structure where markers are organized across different pages. This dimensional change allows direct access within each page while reducing the memory footprint of any single data structure.
Solution Approach 2:
The patent implements dynamic loading of marker address pages based on decoding requirements. Rather than maintaining a static complete table in memory, the system dynamically loads only the pages containing markers needed for the current decoding operation, optimizing memory usage while preserving direct access capability for required markers.
Data Source
AI summary
An image processing device which includes an encoding processing unit which generates a data stream in which markers denoting segmentation of image encoding processing are inserted, and a marker information generation unit which generates marker information which is provided with position information denoting a position in the data stream of a selected marker with a predetermined marker thinning-out interval from the markers which are inserted.