NNPF Image Resampling for High-Resolution Bitstream Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-resolution and high-quality images leads to a surge in transmission and storage costs due to the increase in transmitted information or bit rate, necessitating a high-efficiency image compression technology.
Innovation Solution
An image encoding/decoding method that includes resampling an input picture of a neural-network post-filter (NNPF) based on the size information of the current picture, allowing for efficient encoding/decoding of images with different sizes, and generating a bitstream that can be stored and transmitted.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-resolution and high-quality images are transmitted, then image quality is improved, but transmission costs and storage costs increase
Solution Approach 1:
The patent extracts and transmits only the difference information (residuals) between the original high-resolution image and the down-sampled image, rather than transmitting the entire high-resolution image. This allows reconstruction of high-quality images while significantly reducing the amount of transmitted data.
Solution Approach 2:
The patent introduces a new dimension of processing by applying neural network post-filters that operate in the residual domain. The system processes images at multiple resolutions and uses neural networks to predict and compensate for quality losses, effectively adding a processing dimension that enables efficient compression without quality degradation.
2Adaptability or versatility
If the size of input picture and current picture are different, then encoding flexibility is improved, but encoding complexity increases
Solution Approach 1:
The patent performs down-sampling of the current picture before the neural network filtering operation. By pre-processing the current picture to match the resolution of the input picture, the system simplifies subsequent processing steps and enables the use of standard neural network filters without requiring complex resolution-adaptive implementations.
Solution Approach 2:
The patent dynamically changes the resolution parameter of the current picture through down-sampling to match the input picture dimensions. This parameter adjustment enables flexible handling of images with different sizes while maintaining consistent processing pipelines and reducing the complexity of size-matching operations.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An image encoding/decoding method and device are provided. The image decoding method according to the present disclosure may comprise the steps of: acquiring size information of a current picture; and resampling an input picture of a neural-network post-filter (NNPF) on the basis of the size information.