360 Panoramic Video Site Documentation with SLAM Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for documenting construction and infrastructure sites are inefficient, requiring numerous still images that are difficult to catalog and retrieve, and do not effectively utilize 360-degree panoramic images for visualizing site conditions and comparing 'as-built' construction with planned designs.
Innovation Solution
The method involves extracting 360-degree images from panoramic video sequences using simultaneous localization and mapping (SLAM) and structure from motion (SfM) techniques to determine 3D poses and orientations of images relative to each other and to 2D or 3D plans, creating a photographic record that can be used to inspect and document site conditions, aligning 'as-built' models directly with 'as-planned' models without the need for previously registered anchor images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If individual still photos are used to document site conditions, then comprehensive visual coverage can be achieved, but the process becomes manually intensive and retrieval becomes difficult
Solution Approach 1:
The patent transforms discrete still photo capture into a continuous video recording process. The 360-degree panoramic video continuously captures site conditions throughout construction activities, eliminating the need for manual selection and cataloging of individual photos while maintaining comprehensive documentation coverage.
Solution Approach 2:
The patent creates a digital 3D point cloud model that serves as a virtual copy of the physical construction site. This digital replica allows for efficient storage, retrieval, and analysis of site conditions without requiring physical access to numerous individual photographs.
2Productivity
If 360-degree panoramic images are used to provide complete scene views, then time savings are achieved, but alignment with planned designs becomes more complex
Solution Approach 1:
The patent transitions from 2D panoramic images to 3D point cloud representations. By converting the panoramic video data into a three-dimensional spatial model, the system enables more intuitive and accurate alignment with 3D building information models (BIM) and other multi-dimensional design plans.
Solution Approach 2:
The patent introduces a 3D point cloud model as an intermediary representation between the captured panoramic video and the design plans. This intermediate format serves as a common language that facilitates automated feature matching and alignment between as-built conditions and as-planned designs.
3Measurement precision
If anchor images with known 3D pose are used for calibration, then accurate positioning is achieved, but the process requires additional reference images and computational overhead
Solution Approach 1:
The patent enables the panoramic video data to self-calibrate by extracting and matching features directly from the video sequence itself. The system identifies natural feature points within the panoramic video and uses these intrinsic features for positioning, eliminating the need for external anchor images or manual calibration references.
Data Source
AI summary
A 360 panoramic video sequence of an architectural or industrial site includes a plurality of 360 images. 3D poses of the 360 images are determined with respect to one another, and a subset of the images extracted for further processing according to selection criteria. The 3D poses of the extracted images is refined based on determined correspondences of the features in the images, and a 3D point cloud of the site is developed from the extracted images with refined pose estimates. An “as-built” representation of the site captured in the panoramic video sequence is created from the sparse 3D point cloud and aligned to a 2D or 3D plan of the site. Once so aligned, the extracted images may be presented to a user in conjunction with their plan positions (e.g., overlayed on the site plan). Optionally, a point cloud or mesh view of the site may also be returned.