Hybrid Video Segmentation Using Optical Flow for Moving Objects

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional machine learning-based segmentation techniques fail to produce reliable and consistent segmentation masks in scenarios involving high resolution images, dynamic scenes with fast-moving objects, color-graded contents, dark scenes, and multiple target objects, leading to inconsistent edge clarity and incorrect sub-pixel inference.

Innovation Solution

A hybrid segmentation and optical flow technique combined with gamma correction is applied to adjust brightness and contrast, generating accurate segmentation masks by stacking optical flow maps with segmentation masks, with iterative adjustments to ensure alignment.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If hybrid video segmentation is used to improve segmentation accuracy for objects in motion, then measurement precision is improved, but device complexity increases due to integration of multiple modules

Engineering Contradiction:
Improvesegmentation accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system divides video segmentation into multiple specialized modules: a first segmentation module for stationary objects using background subtraction, and a second segmentation module for moving objects using optical flow analysis. This segmentation allows each module to optimize for its specific task, improving overall accuracy while managing complexity through functional decomposition

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces an optical flow module as an intermediary that processes video frames to detect motion patterns. This intermediary module bridges the gap between raw video input and the second segmentation module, enabling accurate tracking of moving objects without requiring the entire system to handle all segmentation tasks simultaneously

Inventive Principle:
Principle #24Intermediary (Mediator)

2Reliability

If optical flow module is integrated to track moving objects, then reliability of object tracking is improved, but use of energy increases due to additional processing

Engineering Contradiction:
Improveobject tracking reliabilityVSAvoidprocessing energy consumption
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The system dynamically switches between segmentation strategies based on object motion state. The optical flow module is activated specifically for detecting and tracking moving objects, while stationary objects are handled by the more energy-efficient background subtraction method. This dynamic adaptation optimizes energy consumption by applying computational resources only where necessary

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

Different processing quality and complexity are applied to different regions of the video stream based on motion detection. Areas with moving objects receive intensive optical flow processing for reliable tracking, while static regions use simpler background modeling. This local differentiation maintains tracking reliability for moving objects while reducing overall energy consumption

Inventive Principle:
Principle #3Local quality

3Adaptability or versatility

If multiple segmentation modules are used to handle different object types, then adaptability to various scenarios is improved, but device complexity increases

Engineering Contradiction:
Improvescenario adaptabilityVSAvoidmodule integration complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system employs a unified video processing framework that accommodates multiple segmentation strategies. Both the background subtraction module and optical flow module operate within the same processing pipeline, allowing the system to adapt to different scenarios (stationary or moving objects) without requiring separate dedicated systems. This multi-functional architecture improves versatility while controlling complexity through shared infrastructure

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP4154217B1Hybrid video segmentation aided by optical flow for objects in motion
Publication Date: 2026.05.06 SONY GROUP CORP
  • EP4154217B1 patent drawingFigure 1
  • EP4154217B1 patent drawingFigure 2
  • EP4154217B1 patent drawingFigure 3

AI summary

Methods and systems including: applying a gamma adjustment to frames of an input video with particular values of parameters to generate at least one set of gamma-adjusted frames to generate at least one set of gamma-adjusted frames; applying a segmentation technique to the at least one set of gamma-adjusted frames to generate segmentation masks; applying an optical flow technique to the at least one set of the gamma-adjusted frames to generate optical flow maps; and combining the segmentation masks and the optical flow maps to generate hybrid segmentation masks.