Image processing method, training method, system, device, product, and storage medium
By introducing spatial masking into AI image encoding and decoding technology to control the complexity of image encoding and decoding and dynamically adjusting the encoding process, the problem of excessively long image processing time is solved, and more efficient image processing is achieved.
WO2026113613A1PCT designated stage Publication Date: 2026-06-04CLOUD INTELLIGENCE ASSETS HOLDING (SINGAPORE) PTE LTD +1
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- CLOUD INTELLIGENCE ASSETS HOLDING (SINGAPORE) PTE LTD
- Filing Date
- 2025-09-24
- Publication Date
- 2026-06-04
AI Technical Summary
Technical Problem
Existing AI image encoding and decoding technologies have long image processing times due to their complex model structures and high computational requirements, which limits their adoption in real-time applications.
Method used
By introducing spatial masks to control the complexity of image encoding and decoding, the spatial position of mask convolution is generated using complexity information, and the encoding complexity is dynamically adjusted to reduce image processing time.
Benefits of technology
It enables dynamic adjustment of image processing time under different complexity conditions, reducing encoding and decoding time and improving image processing efficiency.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN2025123767_04062026_PF_FP_ABST
Abstract
The present disclosure provides an image processing method, a training method, a system, a device, a product, and a storage medium, applied to the field of image encoding and decoding. The image processing method comprises: acquiring a source image to be encoded; acquiring complexity information, wherein the complexity information represents an image processing duration of the source image; inputting the source image into a trained image encoding system for encoding processing to obtain a hyper-prior bitstream and an entropy-encoded bitstream, and outputting same, wherein the image encoding system is used for encoding the hyper-prior representation of the source image into a hyper-prior bitstream comprising a first spatial mask; and encoding the latent representation of the source image into an entropy-encoded bitstream on the basis of a second spatial mask, wherein the first spatial mask is extracted from the hyper-prior representation on the basis of the complexity information, and is used for representing whether mask convolution needs to be performed at each spatial position in the hyper-prior representation, and the second spatial mask is extracted from the latent representation on the basis of the complexity information, and is used for representing whether mask convolution needs to be performed at each spatial position in the latent representation.
Need to check novelty before this filing date? Find Prior Art