Neural Network Image Compression via Tiled Partitioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image and video compression methods face challenges in decoding large images or videos that exceed memory limits, leading to failed decoding processes.
Innovation Solution
A neural network-based adaptive image and video compression method using a tiled partitioning scheme, which allows for successful decoding by partitioning the image or video into tiles, enabling efficient memory management and processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If the image or video sequence is decoded using conventional methods, then the decoding process can handle standard-sized content, but it fails when the content size exceeds memory limits
Solution Approach 1:
The patent divides the image or video sequence into multiple smaller tiles or blocks that can be independently decoded. Each tile is processed separately within available memory constraints, then reconstructed to form the complete decoded output. This segmentation allows decoding of large images/videos that would otherwise exceed memory capacity.
2Quantity of substance
If a tiled partitioning scheme is applied to enable decoding of large images, then memory management becomes efficient, but the device complexity increases
Solution Approach 1:
The decoding system is segmented into multiple processing units that handle different tiles independently. Each unit contains the necessary decoding functionality to process its assigned tile, enabling parallel processing and efficient memory utilization while distributing system complexity across multiple manageable components.
Solution Approach 2:
The patent introduces a spatial dimension to the decoding process by organizing tiles in a grid layout with specific tile indices and positions. This dimensional organization allows systematic memory management and processing scheduling, transforming a single complex decoding task into multiple structured sub-tasks that can be managed efficiently.
Data Source
AI summary
An image decoding method including transforming an input image into latent samples using an analysis transform; quantizing the latent samples using a hyper encoder to generate quantized hyper latent samples; encoding the quantized hyper latent samples into a bitstream using entropy encoding; applying a latent sample prediction process to obtain quantized latent samples and quantized residual latent samples based on the latent samples using the quantized hyper latent samples; obtaining prediction samples following the latent sample prediction process; and entropy encoding the quantized hyper latent samples and the quantized residual latent samples into the bitstream.