Sub-pixel value generation method in inter prediction and electronic device
By loading multiple reference pixels at once during inter-frame prediction and constructing interpolation operators for multiplication and addition calculations, the high computational complexity of subpixel generation in existing technologies is solved, achieving efficient interpolation calculations and image processing.
CN121357329BActive Publication Date: 2026-07-21BEIJING SANKUAI ONLINE TECH CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING SANKUAI ONLINE TECH CO LTD
- Filing Date
- 2025-11-03
- Publication Date
- 2026-07-21
AI Technical Summary
Technical Problem
Existing sub-pixel generation methods in inter-frame prediction have high computational complexity and low efficiency, and there is a need to improve computational efficiency.
Method used
By loading reference pixels corresponding to M target sub-pixel positions at once, multiple interpolation operators are constructed and multiplication and addition calculations are performed, reducing the number of pixel loadings and improving interpolation calculation efficiency.
Benefits of technology
This greatly improves the efficiency of interpolation calculation, increasing it by M times, thereby improving image processing efficiency.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN121357329B_ABST
Abstract
The present disclosure provides a sub-pixel value generation method in inter prediction and an electronic device. The method comprises: determining a plurality of integer pixel compensation regions of a current prediction unit in a target image, wherein each integer pixel compensation region comprises one integer pixel position and a plurality of sub-pixel positions; determining a first interpolation calculation direction, an interpolation coefficient sequence and a reference pixel sequence corresponding to a target sub-pixel position in the current prediction unit; loading M+N-1 reference pixels corresponding to the target sub-pixel position in M adjacent integer pixel compensation regions in the first interpolation calculation direction, and forming M interpolation operators according to the interpolation coefficient sequence corresponding to the target sub-pixel position, wherein M is greater than or equal to 2; and performing bitwise multiplication accumulation calculation on the M interpolation operators and the M+N-1 reference pixels to generate M sub-pixel values corresponding to the target sub-pixel position in the M integer pixel compensation regions. The present disclosure can improve the image processing efficiency.
Need to check novelty before this filing date? Find Prior Art
Citation Information
Patent Citations
Fast SIMD implement method of HEVC / H.265 sub pixel interpolation
CN104378641A
Multi-core hybrid storage management method for high efficiency video coding (HEVC) process
CN105263022A