Method and apparatus for encoding and decoding an image
By constructing feature maps and parameterized neural networks, and combining motion and composition information, the problem of high complexity in existing neural network video coding is solved, achieving efficient image and video compression at low bit rates, which is suitable for a variety of application scenarios.
CN122162374APending Publication Date: 2026-06-05ORANGE SA
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ORANGE SA
- Filing Date
- 2024-09-12
- Publication Date
- 2026-06-05
Smart Images
Figure CN122162374A_ABST
Abstract
The invention relates to a method and a device for encoding and decoding a signal with a plurality of samples (Pdn) comprising motion information. The decoding method comprises the steps of: - decoding (E31, FMD) a first set of feature maps (FMd i ); - decoding (E35, NND) a set of parameters (Wd k ) representative of a neural network (MLP'); - for at least one sample of the signal to be decoded associated with a position (x n , y n ) in the signal to be decoded, called current sample (Pd n ): - constructing (E33, E34, XTR', TT') a feature vector (Zd n ) from the feature maps of said first set (FMd i ), and: - processing (E36, MLP') said feature vector (Zd n ) using the artificial neural network (MLP') defined by the decoded parameters (Wd k ) in order to provide a vector representative of a decoded value of the current sample (Pd n ).
Need to check novelty before this filing date? Find Prior Art