Plane Estimation from Object Contact Features in Overlapping Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for plane estimation in extended reality environments lead to inaccurate placement of virtual objects, resulting in non-immersive experiences and potential user discomfort due to incorrect depth estimation.
Innovation Solution
A system and method that utilizes at least two overlapping images to detect objects in contact with a plane, identify shared features, determine their poses based on camera poses and disparities, and estimate the plane accurately, enabling precise virtual object placement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional equipment and techniques are used for plane estimation, then the system is simple to operate, but the plane estimation accuracy deteriorates
Solution Approach 1:
The patent transitions from 2D image plane analysis to 3D spatial reasoning by detecting object poses and using camera pose information to reconstruct plane geometry in three-dimensional space. This dimensional transformation enables accurate depth estimation and plane orientation determination that cannot be achieved through conventional 2D image processing alone.
Solution Approach 2:
The patent introduces detected objects as intermediary elements to estimate plane characteristics. Instead of directly analyzing the plane surface, the system detects objects in contact with or near the plane, determines their poses, and uses these object poses as intermediaries to infer the plane's orientation and position, thereby achieving accurate plane estimation through indirect measurement.
2Reliability
If conventional plane estimation methods are used, then the processing time is short, but the visual experience quality deteriorates
Solution Approach 1:
The patent performs preliminary object detection and pose determination using pre-trained detection models and algorithms. By pre-processing the image data to identify objects and their characteristics before plane estimation, the system establishes a foundation for accurate plane reconstruction while optimizing the overall processing workflow to maintain efficiency.
Solution Approach 2:
The patent replaces conventional mechanical or manual plane measurement methods with computational image processing and machine learning-based object detection. This substitution of computational algorithms for traditional measurement techniques enables rapid automated plane estimation while maintaining high accuracy, thereby improving visual experience quality without significant time loss.
3Measurement precision
If inaccurate plane estimation is used, then the system complexity is low, but the virtual object placement accuracy deteriorates
Solution Approach 1:
The patent implements a feedback mechanism where detected object poses and camera pose information are continuously used to refine plane estimation. The system uses the spatial relationships between detected objects and the camera to iteratively improve plane orientation and position accuracy, ensuring precise virtual object placement through continuous validation and adjustment based on detected features.
Data Source
AI summary
Disclosed is system (100) comprising at least one server (102) configured to: obtain at least two images of real-world environment whose fields of view overlap at least partially; obtain pose information indicative of corresponding camera poses from which at least two images are captured; detect, in at least two images, at least one object (202A-202G) that is in contact with given plane (204, 208) present in real-world environment; identify, in at least two images, same features (206A-206J) of at least one object that lie on given plane; determine poses of same features, wherein pose of given same feature is determined, based on disparity in two-dimensional positions of given same feature in at least two images and corresponding camera poses from which at least two images are captured; and estimate given plane based on poses of same features.