Decoding Compressed Digital Images Using Restart Markers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current decoding-decompression methods for digital images and video sequences, particularly those using entropic variable length coding, are susceptible to errors during transmission, leading to significant degradation in image quality due to loss of synchronization and incorrect interpretation of code words.
Innovation Solution
A robust decoding-decompression method that identifies errors by calculating the number of encoded data structures between restart markers and discards or replaces affected data groups, using concealment strategies to maintain image quality, such as motion-compensated replacement in video sequences and spatial interpolation for still images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If conventional decoding-decompression methods using entropic variable length coding are used, then compression efficiency is improved, but error robustness deteriorates
Solution Approach 1:
The bit sequence is divided into data groups separated by restart markers, allowing error confinement to specific segments. Each data group can be independently decoded and error-prone groups can be discarded without affecting the entire sequence, thus maintaining compression efficiency while improving error robustness.
Solution Approach 2:
The patent implements error detection mechanisms that monitor the decoding process and provide feedback when errors are detected. This allows the system to identify corrupted data groups and trigger appropriate error handling procedures, maintaining reliability without sacrificing compression efficiency.
2Reliability
If error detection and concealment procedures are added to improve error robustness, then reliability is improved, but device complexity increases
Solution Approach 1:
Restart markers are pre-inserted into the compressed bit sequence at predetermined intervals before transmission. These markers establish a framework that enables straightforward error detection and concealment during decoding, improving reliability without requiring complex real-time error handling mechanisms.
Solution Approach 2:
The patent employs simple error concealment strategies such as discarding corrupted data groups or using basic interpolation methods. These lightweight approaches provide adequate error robustness without introducing significant computational complexity, effectively treating error handling as a disposable, low-cost operation.
Data Source
AI summary
Presented is a method for decoding-decompressing a compressed-encoded digital data sequence relating to at least one initial digital image. The method includes receiving the digital data sequence having compressed-encoded data groups separated from one another by at least one restart marker and each one including a respective set of encoded data structures. The method calculates a representative value of the number of encoded data structures being between a first restart marker and a subsequent second restart marker signaling, respectively, the start of a first data group to be decoded and the start of a second data group. The method then extracts from the first data group the encoded data structures, and detects the presence of at least one error, if the number of the encoded data structures extracted is different from the calculated value.