Adaptive Media Compression Block Decorrelation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital media compression algorithms, such as the original JPEG standard, fail to provide superior compression performance and quality at comparable bit rates, and modern formats face challenges in gaining widespread acceptance due to computational demands, lack of significant performance improvement, and licensing issues.
Innovation Solution
The development of methods and systems for adaptive compression, recompression, and decompression of digital media using block decorrelating algorithms, which transform and reorganize data to concentrate correlated information, allowing for lossless or lossy compression with improved compression performance, and the ability to transcode between formats without full decompression.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of substance
If modern compression formats (JPEG-2000, JPEG-XR) are used to improve compression performance, then compression ratio is improved, but device complexity and computational demands increase
Solution Approach 1:
The patent divides the image into multiple blocks and processes each block independently through the compression pipeline. This segmentation allows parallel processing and reduces the computational complexity per block, while still achieving overall improved compression ratios through the advanced transform and prediction techniques applied to each segment.
Solution Approach 2:
The patent employs adaptive prediction modes and dynamic mode selection for each block, allowing the compression algorithm to dynamically choose the most efficient processing path based on local image characteristics. This dynamic adaptation improves compression ratios without requiring uniformly high computational resources across the entire image.
2Loss of substance
If advanced compression algorithms are implemented to improve compression performance, then compression ratio is improved, but ease of operation decreases due to licensing fees and format compatibility issues
Solution Approach 1:
The patent implements a universal compression framework that can process images in both progressive and non-progressive modes, supporting multiple prediction techniques and transform types within a single algorithm. This multi-functionality allows the system to adapt to different application requirements and maintain compatibility with various deployment scenarios without requiring multiple separate formats.
Solution Approach 2:
The patent uses computationally intensive advanced prediction and transform techniques only during the compression phase, while the decompression process remains simple and fast. This approach allows the use of sophisticated algorithms where they provide maximum benefit (during encoding) while maintaining ease of operation during decoding and playback.
3Quantity of substance
If lossy compression is applied to reduce file size, then quantity of data is reduced, but measurement precision deteriorates due to quality loss
Solution Approach 1:
The patent applies different prediction modes and compression strengths to different blocks based on their local characteristics. Important regions with edges or high-frequency content receive more careful processing with stronger prediction, while smooth regions can tolerate more aggressive compression. This local quality adaptation reduces overall file size while preserving image quality where it matters most.
Solution Approach 2:
The patent incorporates quality assessment and adaptive quantization where the compression process monitors the impact of compression on image quality and adjusts the compression strength accordingly. This feedback mechanism allows the system to achieve lower file sizes while maintaining acceptable quality levels by dynamically adjusting compression parameters based on actual quality impact.
Data Source
AI summary
Adaptive methods and apparatuses include compressing, recompressing, decompressing, and transmitting/storing digitized media data, such as text, audio, image, and video. Methods may include partitioning data; transforming partitioned data; analyzing partitioned data; organizing partitioned data, predicting partitioned data; partially or fully encoding partitioned data partially or fully decoding partitioned data, and partially or fully restructuring the original data.