Adaptive Threshold Moving Object Detection in Video Sequences

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for detecting moving objects in video image sequences, particularly in scenarios with changing camera angles or lighting conditions, suffer from inaccuracies and high false alarm rates due to image registration issues and temporal changes.

Innovation Solution

An adaptive image change threshold method is implemented, where initial image change points are determined and then revised based on quality criteria, allowing for iterative adjustments to improve robustness and reduce false alarms by selecting prominent feature points and displacement vectors for accurate transformation and difference image analysis.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If image registration is performed to compensate for camera movement and viewing angle changes, then the detection accuracy is improved, but significant inaccuracies still occur during the transformation process

Engineering Contradiction:
Improvedetection accuracyVSAvoidregistration reliability
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent implements an iterative feedback mechanism where the image change threshold is continuously adjusted based on the distribution of image change points. The threshold determination unit repeatedly refines the threshold until the number of image change points falls within the expected range, creating a closed-loop system that automatically corrects for registration inaccuracies and adapts to varying image conditions.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The patent transforms the static image change threshold into a dynamic parameter that automatically adapts to different imaging conditions. By making the threshold flexible and adjustable based on actual image content and change point distribution, the system can respond to varying lighting, camera movement, and scene complexity without requiring manual recalibration.

Inventive Principle:
Principle #15Dynamics

2Ease of operation

If a fixed image change threshold is used to identify image change points, then the processing is simple, but the false alarm rate increases under varying lighting and recording conditions

Engineering Contradiction:
Improveprocessing simplicityVSAvoiddetection reliability
Core Design Contradiction:
Ease of operationVSReliability

Solution Approach 1:

The patent converts the fixed threshold into a dynamic adaptive threshold that automatically adjusts to different imaging conditions. The threshold determination unit calculates and refines the threshold based on the actual distribution of image change points, ensuring optimal detection performance across varying lighting, contrast, and scene conditions without sacrificing processing efficiency.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system performs self-calibration by automatically determining and refining the image change threshold based on the image content itself. The threshold adjustment process is autonomous, using the distribution of image change points to self-correct and optimize detection parameters without external intervention, thereby maintaining both simplicity and reliability.

Inventive Principle:
Principle #25Self-service

3Measurement precision

If the image change threshold is lowered to detect more subtle changes, then more moving objects are detected, but the number of false alarms increases

Engineering Contradiction:
Improvedetection sensitivityVSAvoidfalse alarms
Core Design Contradiction:
Measurement precisionVSObject-generated harmful factors

Solution Approach 1:

The patent uses feedback control to dynamically adjust the threshold based on the number and distribution of detected image change points. When too many change points are detected (indicating potential false alarms), the threshold is automatically raised; when too few are detected (missing actual objects), the threshold is lowered. This closed-loop mechanism maintains optimal sensitivity while minimizing false alarms.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The patent dynamically changes the image change threshold parameter based on the detected image content and change point distribution. By adjusting this critical parameter adaptively rather than using a fixed value, the system optimizes the balance between detection sensitivity and false alarm rate for each specific imaging scenario.

Inventive Principle:
Principle #35Parameter changes

4Reliability

If multiple iterations of threshold adjustment are performed to improve detection quality, then the false alarm rate is reduced, but the processing time increases

Engineering Contradiction:
Improvedetection qualityVSAvoidprocessing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent implements a stopping criterion that prevents excessive iterations by terminating the threshold adjustment process when the number of image change points falls within the expected range. This ensures that sufficient iterations are performed to achieve reliable detection while avoiding unnecessary additional iterations that would only increase processing time without improving detection quality.

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The iterative threshold adjustment process incorporates feedback-based termination conditions that automatically stop iterations when detection quality reaches an acceptable level. The system monitors the number of image change points and halts the refinement process when this count falls within the expected range, optimizing the trade-off between detection reliability and processing efficiency.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP2590141B1Method and device for detecting moving objects in a video picture sequence
Publication Date: 2019.01.02 AIRBUS DEFENCE & SPACE GMBH
  • EP2590141B1 patent drawingFigure 1
  • EP2590141B1 patent drawingFigure 2~3
  • EP2590141B1 patent drawingFigure 4~5

AI summary

The method involves establishing an image variation threshold value and determining image variation points in difference image with an absolute image brightness value that exceeds the image variation threshold value. The quality of image variation points Is analyzed based on at least one predetermined quality criterion. The image variation points are established as distinctive image variation points with different image variation threshold value if the quality criterion is met. An independent claim is included for a device for the detection of moving objects in a video Image sequence.