Neural Upsampling of Correlated Multichannel Data After Lossy Compression

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing data compression methods for complex-valued SAR images are inadequate as they are prone to loss of information, leading to compression artifacts that affect interpretability, and are limited in real-time processing for applications like disaster response and surveillance.

Innovation Solution

A system and method for upsampling decompressed correlated multichannel data after lossy compression using a neural network, which integrates AI-based techniques to enhance compression quality. This involves a novel AI deblocking network with recurrent layers for feature extraction and a channel-wise transformer with attention to capture complex inter-channel dependencies.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If lossy compression is applied to SAR images to reduce bandwidth and storage requirements, then compression ratio is improved, but information loss and compression artifacts increase

Engineering Contradiction:

Improvedata sizeVSAvoidinformation loss

Core Design Contradiction:

Quantity of substanceVSLoss of information

Solution Approach 1:

A neural network up-sampler is introduced as an intermediary component between the lossy compression decoder and the final SAR image output. The up-sampler takes the down-sampled complex-valued SAR image from the decoder and reconstructs it to the original resolution, acting as a mediator that recovers information lost during compression while maintaining the benefits of lossy compression

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The neural network up-sampler is pre-trained on pairs of down-sampled and original SAR images to learn the mapping relationship. This preliminary training enables the up-sampler to automatically recover lost information during the decompression process without requiring additional computational resources during real-time operation

Inventive Principle:

Principle #10Preliminary action

2Ease of operation

If conventional compression algorithms are used for SAR images, then ease of operation is improved, but manufacturing precision and information recovery deteriorate

Engineering Contradiction:

Improvecompression simplicityVSAvoidimage quality

Core Design Contradiction:

Ease of operationVSManufacturing precision

Solution Approach 1:

The invention merges conventional lossy compression algorithms with a neural network up-sampler into a hybrid compression system. The conventional algorithm handles the initial compression to achieve high compression ratios, while the neural network up-sampler is integrated to restore image quality, combining the advantages of both approaches

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The system allows dynamic adjustment of compression parameters including compression ratio, down-sampling factor, and neural network architecture parameters. This enables optimization of the balance between compression efficiency and image quality recovery based on specific application requirements

Inventive Principle:

Principle #35Parameter changes

3Loss of information

If deep neural networks are used for SAR image compression to reduce information loss, then information recovery is improved, but device complexity and processing time increase

Engineering Contradiction:

Improveinformation recoveryVSAvoidsystem complexity

Core Design Contradiction:

Loss of informationVSDevice complexity

Solution Approach 1:

The compression system is segmented into two independent components: a conventional lossy compression module and a neural network up-sampler module. This segmentation allows each component to be optimized separately - the conventional module for compression efficiency and the neural network module for quality recovery - reducing overall system complexity compared to using a single deep neural network for the entire compression process

Inventive Principle:

Principle #1Segmentation

4Productivity

If lossy compression is applied to complex-valued SAR images, then productivity is improved, but measurement precision and interpretability deteriorate

Engineering Contradiction:

Improveprocessing speedVSAvoidphase accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The neural network up-sampler applies localized processing to recover phase information in different regions of the SAR image. By focusing computational resources on recovering phase accuracy in critical regions while maintaining speed in less critical areas, the system achieves a balance between measurement precision and processing speed

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS20250192798A1System and methods for upsampling of decompressed correlated multichannel data using a neural network

Publication Date: 2025.06.12 ATOMBEAM TECH INC

AI summary

A system and methods for upsampling of decompressed correlated multichannel data after lossy compression using a neural network that integrates AI-based techniques to enhance compression quality. It incorporates a novel AI deblocking network composed of recurrent layers for feature extraction and a channel-wise transformer with attention to capture complex inter-channel dependencies. The recurrent layers extract multi-dimensional features from the two or more correlated datasets, while the channel-wise transformer learns global inter-channel relationships. This hybrid approach addresses both local and global features, mitigating compression artifacts and improving decompressed data quality. The model's outputs enable effective data reconstruction, achieving advanced compression while preserving crucial information for accurate analysis.