Foreground Extraction Using Dynamic Background Update Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques struggle to accurately detect and extract a foreground from images where objects, such as people, remain motionless for a short period or move in a way that prevents their detection as a foreground, leading to erroneous detection or failure in extraction.
Innovation Solution
An image processing apparatus that generates a background image by determining motionless pixels across multiple frames and allows user-controlled settings for background image updates, enabling proper foreground extraction by comparing input images with the generated background.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If background differencing is used to extract foreground by detecting motion, then moving objects can be detected, but motionless objects or objects moving below detection threshold cannot be properly extracted
Solution Approach 1:
The system dynamically adjusts the background image update strategy based on detection confidence. When an object becomes motionless or moves minimally, the system suppresses automatic background updates in that region, maintaining the object as foreground. This dynamic adaptation allows the system to handle both moving and motionless objects effectively.
Solution Approach 2:
The system uses detection results as feedback to control background image updates. When detection confidence is high, normal background updates occur. When confidence is low (indicating a potentially motionless object), the system adjusts update behavior to prevent misclassification, creating a feedback loop that improves detection accuracy for various object states.
2Reliability
If the background image is continuously updated to reflect current scene, then background accuracy improves, but motionless objects may be incorrectly classified as background
Solution Approach 1:
The system performs preliminary detection to identify objects with low motion activity before updating the background image. By detecting potential motionless objects in advance and suppressing background updates for those regions, the system prevents misclassification while still allowing normal background updates in active regions.
Solution Approach 2:
The background update operation is applied locally rather than globally. The system determines update suppression on a per-region basis, allowing background updates in regions with moving objects while preventing updates in regions with motionless objects. This local differentiation maintains overall system reliability.
3Productivity
If motion-based detection is used, then moving foreground objects can be extracted, but objects with minimal or no motion are missed
Solution Approach 1:
The system introduces an intermediary detection mechanism that operates between simple motion detection and complex full-frame analysis. By using detection confidence as an intermediary metric, the system can quickly identify potential motionless objects and apply targeted background update suppression without performing computationally intensive analysis on every frame.
Data Source
AI summary
An image processing apparatus extracts a foreground image corresponding to an object included in a processing image using a background image corresponding to the processing image, and generates the background image from the processing image. The image processing apparatus determines whether it is allowed to update the background image for use in the extraction, and based on a result of the determination, updates the background image for use in the extraction using the generated background image.