Background Model Segmentation for Video Surveillance Adaptation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing background subtraction algorithms face challenges in balancing the stability of distinguishing between background and non-background while quickly adjusting to changes in lighting conditions, background objects, and camera orientation, requiring a method to create a stable background that adapts to these changes.
Innovation Solution
An image processing device and method that create multiple background models based on the statistical distribution of image frame sets of varying lengths, using a background model creation section to analyze pixel value histograms and a background image creation section to determine pixel values from these models, allowing for the creation of a stable background image that adjusts to changes by referencing multiple models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Stability of the object's composition
If observation of input images is performed over a long period of time to stably distinguish between background and non-background, then stability in distinguishing background is improved, but ability to adjust to background changes deteriorates
Solution Approach 1:
The patent divides the background model into multiple segments corresponding to different time periods (first time period and second time period). Each segment is updated independently based on its respective time window, allowing the system to maintain historical stability information while incorporating recent changes. This segmentation enables simultaneous access to long-term statistical stability and short-term adaptability.
Solution Approach 2:
The patent implements dynamic updating of background models where the second background model (representing recent time period) is continuously updated with new input images, while the first background model (representing historical time period) maintains longer-term stability. The system dynamically adjusts between these models based on current conditions, enabling adaptation to changing environments while preserving stable background representation.
2Adaptability or versatility
If multiple background models based on different time periods are created, then ability to adjust to background changes is improved, but device complexity increases
Solution Approach 1:
The patent employs periodic updating mechanisms where the background models are updated at different rates corresponding to different time periods. The first background model updates based on historical data with longer periodicity, while the second background model updates more frequently based on recent data. This periodic action structure simplifies management by establishing regular update cycles rather than requiring complex continuous adjustment mechanisms.
Solution Approach 2:
The patent introduces a background model management mechanism that acts as an intermediary between multiple background models and the background subtraction process. This intermediary selectively references appropriate background models based on current conditions, managing the complexity of having multiple models by providing a unified interface and decision-making layer that determines which model to use or how to combine them.
Data Source
AI summary
Disclosed herein is an image processing device including a background model creation section adapted to create a plurality of background models based on statistical distribution of a plurality of image frame sets, the image frame sets differing in length of time, and each of the image frame sets including a plurality of input frames; and a background image creation section adapted to create a background image by referring to the plurality of background models.