3D Reconstruction Scale Factor Using Camera Base Reference Points
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 3D reconstruction methods using 360-degree cameras cannot accurately determine the real-world dimensions of a reconstructed 3D space, leading to arbitrary scale measurements that hinder measurement and comparison applications.
Innovation Solution
A method and apparatus that utilize a camera with a supporting base and reference points to capture images, apply deep learning for depth estimation, and calculate a scale factor based on identified 3D points and angles to convert arbitrary dimensions into real-world dimensions, eliminating the need for depth sensors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If 360-degree cameras are used for 3D reconstruction, then the ability to capture and reconstruct space geometry is improved, but the measurement precision of real-world dimensions deteriorates because the dimensions are only in arbitrary scale
Solution Approach 1:
A scale factor is introduced as an intermediary parameter to bridge the gap between arbitrary 3D reconstruction dimensions and real-world dimensions. The scale factor is calculated using the known distance between reference points on the supporting base, allowing conversion of virtual 3D measurements into accurate real-world measurements without requiring depth sensors during reconstruction
Solution Approach 2:
The system changes the scale parameter of the 3D reconstruction by calculating a scale factor that transforms arbitrary units into real-world units. This parameter change is achieved by comparing the known physical distance between reference points with their corresponding distance in the 3D reconstructed space, thereby establishing an accurate scaling relationship
2Ease of operation
If learned depth-prediction methods are used, then the ease of operation is improved, but the measurement precision deteriorates due to depth estimation errors and scale inaccuracies
Solution Approach 1:
The patent replaces learned depth-prediction methods with a geometric calculation approach. Instead of using neural networks to estimate depth, the system calculates the scale factor using known reference point distances and basic geometric relationships, eliminating the depth estimation errors inherent in learned methods while maintaining ease of operation
Solution Approach 2:
The system creates a scaled copy of the real-world environment in 3D space and then applies a calculated scale factor to transform this copy into accurate real-world dimensions. This copying approach with subsequent scaling is more precise than direct depth prediction while remaining computationally efficient
3Adaptability or versatility
If multiple shots of 360 camera are used to estimate scene geometry, then the adaptability is improved, but the device complexity increases due to the need for multiple captures and processing
Solution Approach 1:
The supporting base with reference points serves a dual function: it stabilizes the camera during capture and provides the reference measurements needed for scale factor calculation. This self-service approach eliminates the need for separate depth sensors or complex calibration equipment, reducing device complexity while maintaining accurate geometry estimation
Data Source
AI summary
A method (700) of determining a scale factor is provided. The method comprises capturing (s702) a real-world environment using a camera resting on a plane, thereby generating an image. The camera comprises a supporting base including a first reference point. The method further comprises, based on the image, identifying (s704) a first three-dimensional (3D) point of a virtual 3D environment that is a reconstruction of the real-world environment. The first 3D point is mapped to the first reference point of the supporting base. The method further comprises determining (s706) the scale factor based on a coordinate of the first 3D point.