Background Image Updating to Exclude Stationary Foreground Objects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for generating background images fail to accurately separate foreground objects that remain stationary for an extended period, leading to their inclusion in the background image, which hinders faithful reproduction of 3D models in virtual viewpoint images.
Innovation Solution
An image processing apparatus that generates and updates background images by identifying updating and non-updating areas using area information, employing multiple captured images to prevent stationary objects from being incorporated into the background.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If the background image is generated by combining multiple captured images during a predetermined time, then the background image can be constructed from available data, but stationary objects are incorrectly included in the background image
Solution Approach 1:
The patent divides the background image into multiple regions (first region and second region) based on temporal characteristics. The first region corresponds to areas where background images are updated using captured images, while the second region corresponds to areas where background images are not updated. This segmentation allows the system to handle different areas differently, preventing stationary objects from being incorrectly included in the background image while still constructing the background image from available data.
Solution Approach 2:
The patent applies different processing qualities to different regions of the background image. In the first region, the background image is updated using captured images to reflect current scene information. In the second region, the background image is not updated to preserve historical information and avoid incorporating stationary objects. This local differentiation of processing quality resolves the contradiction between using available data and maintaining accurate object separation.
2Adaptability or versatility
If the background image is updated using captured images, then the background image reflects current scene information, but stationary objects remain in the background image
Solution Approach 1:
The patent segments the background image into updated regions and non-updated regions. By applying update operations only to the first region and excluding the second region from updates, the system maintains adaptability to current scene information while preventing the incorporation of stationary objects into the background image, thus improving foreground-background separation accuracy.
Solution Approach 2:
Instead of updating the entire background image uniformly, the patent inverts the approach by selectively applying updates only to specific regions while excluding other regions. This inversion of the conventional uniform update approach allows the system to maintain background image relevance while avoiding the inclusion of stationary objects.
3Measurement precision
If area information is used to distinguish updating and non-updating areas, then stationary objects can be excluded from the background image, but the system complexity increases
Solution Approach 1:
The patent uses area information to segment the background image into distinct regions (first region for updating, second region for non-updating). This segmentation approach enables precise control over which areas are updated and which are not, achieving accurate object separation while managing system complexity through structured region management.
Solution Approach 2:
The patent prepares area information in advance to identify which regions should be updated and which should not be updated. By pre-determining the updating and non-updating areas based on area information, the system can efficiently process captured images without requiring complex real-time analysis, thus achieving accurate object separation while controlling system complexity.
Data Source
AI summary
In a case where a background image is generated from a plurality of captured images obtained during a predetermined time, on a condition that an object at rest exists, the object is taken in in the background image. A new background image is generated by performing updating processing based on the plurality of captured images, which takes an updating area identified by area information among generated background images corresponding to the plurality of captured images obtained during a predetermined time as a target. The area information is information indicating an updating area, which is a target of updating processing, and a non-updating area, which is not a target of updating processing, in the background image.