Abandoned Object Detection via Dynamic Background Modeling

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current image processing technologies for video surveillance systems are inadequate for detecting abandoned objects on roads, as existing methods are either manually intensive, sensitive to noise, or have limitations in adapting to changing light conditions and complex scenes, leading to inefficient and inaccurate identification of illegal road occupations.

Innovation Solution

An abandoned object detection apparatus and method that involves matching each pixel of a current frame with a background model, marking unmatched pixels as foreground, updating the background model, and using a mask processing unit to identify and process abandoned objects, thereby reducing the influence of occlusion and ghost phenomena.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual inspection is used to identify illegal road occupation, then detection accuracy can be maintained, but labor cost and time consumption increase significantly

Engineering Contradiction:

Improvedetection accuracyVSAvoidtime consumption

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent replaces manual mechanical inspection with an automated image processing system that uses background modeling and frame difference algorithms to detect abandoned objects, eliminating human labor while maintaining detection accuracy through computational methods

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system enables self-service detection by automatically analyzing video frames, updating background models, and identifying abandoned objects without human intervention, allowing the system to operate autonomously and reduce time consumption

Inventive Principle:

Principle #25Self-service

2Measurement precision

If optical flow method is used for target detection, then moving target extraction is achieved, but computation complexity increases and noise sensitivity rises

Engineering Contradiction:

Improvetarget extraction accuracyVSAvoidcomputation complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts only the essential features needed for abandoned object detection by using frame difference algorithms that compare pixel values between frames, rather than computing the full optical flow field, thereby reducing computation complexity while maintaining detection capability

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The system applies partial action by performing detection only on regions where changes are detected through frame differencing, rather than processing the entire image with full optical flow computation, reducing overall computational burden

Inventive Principle:

Principle #16Partial or excessive action

3Adaptability or versatility

If frame difference algorithm is used, then dynamic scene handling is improved, but false detection increases due to ghost phenomena

Engineering Contradiction:

Improvedynamic scene adaptabilityVSAvoiddetection reliability

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The patent implements feedback mechanisms by continuously updating the background model based on detected frames and using this updated model for subsequent detections, allowing the system to adapt to changing scenes while reducing false detections through iterative refinement

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system dynamically adapts to changing scenes by updating the background model in real-time and adjusting detection parameters based on current scene characteristics, enabling effective handling of dynamic environments while maintaining detection reliability

Inventive Principle:

Principle #15Dynamics

4Productivity

If background modeling approach is used, then calculation speed is improved, but detection accuracy decreases under complex scenes and varying light conditions

Engineering Contradiction:

Improvecalculation speedVSAvoiddetection accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent employs dynamic background modeling that adapts to changing light conditions and complex scenes by continuously updating the background model and adjusting detection thresholds, maintaining both calculation speed and detection accuracy in varying environments

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system changes detection parameters dynamically based on scene characteristics, adjusting sensitivity thresholds and model update rates according to lighting conditions and scene complexity, thereby maintaining accuracy while preserving computational efficiency

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS10212397B2Abandoned object detection apparatus and method and system

Publication Date: 2019.02.19 FUJITSU LTD

AI summary

An abandoned object detection apparatus and method and a system where the apparatus is configured to match each pixel of an acquired current frame with its background model, mark unmatched pixels, taken as foreground pixels, on a foreground mask, add 1 to a foreground counter to which each foreground pixel corresponds, and update the background model; for each foreground pixel, mark a point corresponding to the foreground pixel on an abandon mask when a value of the foreground counter to which the foreground pixel corresponds is greater than a second threshold value; and for each point on the abandon mask, process the abandon mask according to its background model and buffer background or foreground mask. Hence, when the abandoned object leaves and how long it stays may be judged, and interference of occlusion and ghost may also be avoided, thereby solving a problem of illegal road occupation identification.