Video Background Subtraction Using Multi-Plane Pixel History
Solution Overview
Problem
Existing background subtraction techniques in video analytics face challenges in accurately classifying pixels as background or foreground, especially in dynamic scenes, and are inefficient in terms of processing time and memory usage.
Innovation Solution
The proposed method employs a system that maintains pixel background and foreground histories using multiple history planes, classifying pixels based on differences with stored values and updating these planes periodically, allowing for efficient classification and reduction of noise in the foreground mask.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional background subtraction techniques are used to classify pixels as background or foreground, then the system can detect moving objects, but the accuracy is reduced in dynamic scenes and processing time increases
Solution Approach 1:
The patent segments the pixel classification process into distinct phases by maintaining multiple history planes (e.g., short-term and long-term background histories) that are updated at different rates. This segmentation allows the system to process pixels more efficiently by comparing against appropriate history planes based on scene dynamics, improving accuracy without uniformly increasing processing time for all pixels.
Solution Approach 2:
The patent implements periodic updates of background history planes at controlled intervals rather than continuously updating all planes for every frame. This periodic action reduces the overall processing burden while maintaining accurate background models, thereby decreasing processing time while preserving classification accuracy through strategically timed updates.
2Measurement precision
If multiple history planes are maintained for pixel classification, then the accuracy of background subtraction is improved, but the memory usage increases
Solution Approach 1:
The patent applies local quality by maintaining different types of history planes for different spatial or temporal regions of the image. For example, certain areas with high motion activity may use one history plane configuration while static areas use another, optimizing memory allocation to where it is most needed for accuracy rather than uniformly across the entire image.
Solution Approach 2:
The patent dynamically adjusts parameters of the history planes such as the number of planes, update frequencies, and comparison thresholds based on scene characteristics. This allows the system to optimize the balance between memory usage and accuracy by adapting the history plane configuration to match the actual dynamics of the scene being processed.
3Measurement precision
If pixel values are compared against multiple history planes, then the classification accuracy is improved, but the processing complexity increases
Solution Approach 1:
The patent introduces dynamics by making the selection and updating of history planes adaptive rather than static. The system dynamically determines which history planes to compare against based on current scene conditions, and dynamically updates planes at different rates based on their utility. This dynamic approach manages processing complexity by being selective rather than exhaustively comparing all planes in all conditions.
4Adaptability or versatility
If the background history plane is updated frequently, then the adaptability to scene changes is improved, but the processing time and memory operations increase
Solution Approach 1:
The patent implements periodic updates of background history planes at controlled intervals rather than continuously updating all planes for every frame. This periodic action reduces the overall processing burden while maintaining accurate background models, thereby decreasing processing time while preserving classification accuracy through strategically timed updates.
Solution Approach 2:
The patent introduces dynamics by making the selection and updating of history planes adaptive rather than static. The system dynamically determines which history planes to compare against based on current scene conditions, and dynamically updates planes at different rates based on their utility. This dynamic approach manages processing complexity by being selective rather than exhaustively comparing all planes in all conditions.
Data Source
AI summary
Methods and an apparatus for background subtraction are described. According to some implementations, the temporal history of a pixel location can be used to classify whether a current input pixel of a current input video frame is a background pixel or a foreground pixel. The classification can be made based on the number of matching pixels in corresponding pixel locations in background history planes. One of the background history planes can be updated according to the classified background pixels and persistent classified foreground pixels.