Dynamic Background Model for Object Posture Determination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing imaging technologies face challenges in accurately recognizing the posture of an object in dynamic environments where the background changes or the camera moves, leading to potential errors and limited application range.
Innovation Solution
An imaging apparatus and method that divides an image frame into regions, calculates moving directions and distances, and applies preset background and object models to determine the posture of an object by generating a background model and object model based on these calculations, using a body part model to estimate the posture and predict object and background regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a fixed background model is used to determine object posture, then the method is simple and easy to implement, but the accuracy deteriorates when the background changes or camera moves
Solution Approach 1:
The patent applies dynamics by transitioning from a fixed background model to a dynamic background model that adapts to changing environments. The system continuously updates the background model based on multiple image frames, allowing it to accommodate background changes and camera movements while maintaining posture determination accuracy.
Solution Approach 2:
The patent changes parameters by adjusting the background model parameters dynamically. Instead of using a static background representation, the system modifies background parameters based on observed changes across multiple frames, enabling accurate object posture determination even when the background or camera position changes.
2Device complexity
If the background region is assumed to be motionless, then the processing complexity is reduced, but the adaptability to various environments deteriorates
Solution Approach 1:
The patent applies segmentation by dividing the image into multiple regions and processing each region separately. This allows the system to identify moving objects against changing backgrounds by comparing regional differences across frames, improving environmental adaptability without excessively increasing overall processing complexity.
Solution Approach 2:
The patent applies preliminary action by pre-processing multiple image frames to establish a dynamic background model before determining object posture. This preliminary processing enables the system to adapt to various environments by anticipating and accounting for background changes and camera movements in advance.
3Productivity
If mask information is generated for the background region previously, then the object posture can be determined through matching, but errors increase when the background region changes
Solution Approach 1:
The patent applies feedback by using information from multiple image frames to continuously refine the background model. The system compares current frames with previous frames, uses the differences to update the background representation, and then uses this updated model for posture determination, thereby maintaining reliability even when backgrounds change.
Solution Approach 2:
The patent applies preliminary action by generating and updating mask information for the background region in advance using multiple frames. This preliminary processing creates a more robust background model that accounts for potential changes, improving the reliability of subsequent posture determination through matching.
Data Source
AI summary
An imaging apparatus is provided. The imaging apparatus includes an imager configured to record an external image, a region determiner configured to divide a recorded previous image frame and a recorded current image frame into pluralities of regions, calculate a moving direction and a distance of each of the plurality of regions in the current image frame corresponding to each of the plurality of regions in the previous image frame, and determine a background and an object by applying a preset background model and object model based on the calculated moving direction and distance, and a posture determiner configured to determine body parts of the determined object based on a body part model, and determine a posture of the object by combining the determined body parts.