3D Body Model Reconstruction from 2D Images via Accelerometer Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current mobile devices lack the capability to create and analyze 3D models of the human body using 2D images without requiring a depth map or 3D cameras, which are not commonly available in the mass market, limiting the ability to scan, display, and share accurate 3D body metrics.
Innovation Solution
A method utilizing a single 2D camera on a mobile device to capture multiple 2D images, combined with accelerometer data to determine camera angles, for creating a 3D body model without the need for depth measurements, allowing for automatic segmentation and measurement extraction, and sharing of the 3D model for various applications such as shopping and manufacturing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a 3D camera with depth map capability is used, then measurement precision and 3D model accuracy are improved, but device complexity and cost increase significantly
Solution Approach 1:
The patent creates a digital 3D copy of the human body model from 2D images, allowing virtual representation and measurement without requiring physical 3D scanning hardware. The system generates a computational model that replicates body measurements and morphology, enabling accurate measurements through image processing rather than direct optical measurement.
Solution Approach 2:
The patent replaces the mechanical/optical depth sensing system (3D cameras, infrared sensors, structured light projectors) with a computational approach using 2D image capture and mathematical reconstruction. The mechanical 3D measurement system is substituted with algorithmic processing of 2D images combined with accelerometer data to derive 3D body metrics.
2Measurement precision
If multiple 2D cameras with synchronization hardware are used, then 3D model accuracy is improved, but device complexity and cost increase
Solution Approach 1:
The patent transitions from using multiple cameras viewing the object simultaneously (spatial arrangement) to using a single camera capturing images at different angular positions (temporal/spatial transformation). The system rotates the subject or camera to capture 2D images from multiple angles, then reconstructs 3D information computationally, replacing a multi-camera spatial array with a single camera performing angular sweeps.
Solution Approach 2:
The patent introduces dynamic element rotation or subject rotation to enable a single stationary camera to capture images from multiple angular positions. This dynamic approach replaces the static multi-camera arrangement, using temporal sequences of 2D images taken during rotation to reconstruct 3D body morphology.
3Measurement precision
If infrared sensors and additional hardware are added to create a 3D camera, then depth measurement capability is improved, but ease of manufacture and device accessibility worsen
Solution Approach 1:
The patent makes the 2D camera perform multiple functions: capturing frontal images, capturing lateral images, and providing data for 3D reconstruction. The same standard camera hardware is used for diverse imaging tasks without requiring specialized depth-sensing components, enabling mass-market device compatibility.
Solution Approach 2:
The patent uses inexpensive standard 2D camera sensors that are already ubiquitous in mobile devices, replacing expensive specialized 3D sensing hardware. The approach leverages readily available, low-cost components rather than requiring rare or expensive specialized sensors, dramatically improving accessibility.
4Productivity
If automated segmentation and 3D reconstruction algorithms are implemented, then productivity and measurement extraction speed are improved, but computational complexity and processing requirements increase
Solution Approach 1:
The patent performs automated segmentation of the body from the background and automated identification of anatomical landmarks in advance of measurement extraction. This preliminary processing prepares the data structure beforehand, enabling rapid measurement calculation without requiring complex real-time computation during the measurement phase.
Solution Approach 2:
The system performs automated self-segmentation and self-measurement extraction without requiring manual annotation or intervention. The algorithms automatically identify body contours, landmarks, and dimensions from the captured images, enabling the system to serve itself in the measurement process and reducing operational complexity.
Data Source
AI summary
A 2D camera is used to create a 3D full body image. A camera takes 3 or more 2D images, an accelerometer is used to calculate camera position, and a CPU is employed to construct a 3D body model. This may be performed in a non-controlled environment and by the user alone. An automatic segmentation of the 2D images creates special information for 3D model reconstruction. Once the 3D model measurements are extracted, the user has the option to further specify measurements. In one embodiment, the 3D model is shared via cloud and social media, and also used to assist in shopping while ensuring accurate measurements for the user. In another embodiment, the digital model of products designed for the target consumer body is automatically adjusted and shown as a 3D image on the user's body. The 3D model may be shared with businesses for manufacturing using 3D morphology.