Pixel Value Bins for Moving Object Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for identifying moving objects in a scene with multiple objects in close proximity struggle to effectively generate a background image for change detection processing.
Innovation Solution
A method that initializes pixel value bins to accumulate and average pixel values, determining a stationary pixel value from the bin with the highest number of entries, and compares subsequent pixel values to this average to identify moving objects based on a predetermined threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional change detection methods are used with multiple moving objects in close proximity, then object identification becomes more difficult, but computation time and processing complexity increase significantly
Solution Approach 1:
The patent segments the image processing task by dividing the scene into multiple regions of interest (ROIs) based on detected edges and contours. Each ROI is processed independently to identify moving objects, reducing the overall computational complexity while maintaining identification accuracy for multiple objects in close proximity.
Solution Approach 2:
The patent applies local quality by using adaptive thresholding and local statistics (mean and standard deviation) within each ROI rather than global processing. This allows the system to adapt to local variations in lighting and texture, improving object identification accuracy in complex scenes with multiple objects.
2Productivity
If background image is generated from previous frames using simple averaging, then processing speed is maintained, but accuracy deteriorates when multiple moving objects are present
Solution Approach 1:
The patent performs preliminary actions by detecting edges and contours before generating the background image. This preliminary segmentation allows the system to exclude moving objects from background computation more effectively, improving background accuracy without significantly increasing processing time.
Solution Approach 2:
The patent implements dynamic background generation by adaptively adjusting the blending factor between current and previous background images based on motion detection results. This dynamic adjustment allows the system to maintain processing speed while improving background accuracy in scenes with multiple moving objects.
3Stability of the object's composition
If multiple previous frames are averaged to create background, then background stability improves, but processing time increases
Solution Approach 1:
The patent applies partial action by using a weighted combination of a limited number of previous frames rather than averaging all available frames. This selective approach maintains background stability while reducing processing time by focusing on the most relevant historical frames.
Solution Approach 2:
The patent implements periodic background updates by selectively refreshing the background image based on motion activity levels. When motion is detected, the background is updated more frequently; when the scene is static, updates are reduced, maintaining stability while optimizing processing time.
Data Source
AI summary
A system for identifying pixels associated with moving objects in a series of image frames establishes a stationary pixel value for the portion of the image that is stationary, that is, the background. As the stationary background is generally visible more often than the moving objects, each of the pixel values in a sequence of frames is placed into one of three groups or bins, each bin having a respective range of values. The bin with the most pixel values placed therein represents the stationary background and the average of the pixel values in that bin is set as the background pixel value against which pixel values in subsequent frames are compared to identify the moving objects in the image.