Neural Video Frame Interpolation for FPS Mismatch and Motion Blur

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The mismatch in frames per second (FPS) between image data and the apparatus reproducing it leads to deteriorated image quality, necessitating frame interpolation or synthesis to improve video quality.

Innovation Solution

A display device employs neural networks for motion estimation and object detection to generate interpolation frames by using pre-generated mapping tables and metadata, reducing data processing complexity and increasing FPS through frame rate conversion.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If conventional video processing methods are used, then device complexity is reduced, but video quality deteriorates due to block artifacts and blurriness

Engineering Contradiction:
Improvevideo qualityVSAvoidprocessing complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The video processing method segments the video data into multiple frames and further divides each frame into blocks, then applies different processing operations to different blocks based on their characteristics. This segmentation allows selective processing that improves video quality while managing computational complexity through localized operations rather than global processing.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies different processing operations to different blocks within video frames based on local characteristics such as motion magnitude and block type (intra-block vs inter-block). Motion-compensated interpolation is applied to blocks with high motion magnitude, while simpler operations are used for static blocks, optimizing the balance between quality improvement and processing complexity.

Inventive Principle:
Principle #3Local quality

2Manufacturing precision

If motion-compensated interpolation is applied to all blocks, then video quality improves, but processing time and complexity increase significantly

Engineering Contradiction:
Improveinterpolation accuracyVSAvoidprocessing time
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

Instead of applying motion-compensated interpolation to all blocks, the patent selectively applies this operation only to blocks with high motion magnitude that would benefit most from it. Static or low-motion blocks receive simpler processing, reducing overall processing time while maintaining video quality where it matters most.

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The patent changes the processing parameter (interpolation operation type) based on the motion magnitude parameter of each block. By dynamically adjusting the processing intensity according to local motion characteristics, the system achieves high interpolation accuracy where needed while minimizing processing time for static regions.

Inventive Principle:
Principle #35Parameter changes

3Manufacturing precision

If advanced processing operations are used, then video quality improves, but ease of operation deteriorates due to complex processing requirements

Engineering Contradiction:
Improvevideo qualityVSAvoidprocessing simplicity
Core Design Contradiction:
Manufacturing precisionVSEase of operation

Solution Approach 1:

The video processing system automatically analyzes each block's characteristics (motion magnitude, block type) and selects the appropriate processing operation without requiring manual intervention. The system self-adjusts the processing complexity based on the content being processed, making advanced video quality enhancement accessible without requiring operator expertise in complex processing parameters.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP4184424B1Method and device for improving video quality
Publication Date: 2026.04.29 SAMSUNG ELECTRONICS CO LTD
  • EP4184424B1 patent drawingFigure 1
  • EP4184424B1 patent drawingFigure 2
  • EP4184424B1 patent drawingFigure 3

AI summary

Provided is a method of improving video quality. The method of improving video quality includes obtaining first motion information regarding a user control object from a frame included in a video, obtaining second motion information regarding remaining pixels excluding pixels included in the user control object from pixels included in the frame, by using a first neural network, and generating an interpolation frame between at least two frames from the at least two frames, the first motion information, and the second motion information, by using a second neural network.