Camera Pose Optimization via Stationary Reference Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing geographic positioning techniques often yield inaccurate results when determining the location of objects in a geographic area from photographic imagery, particularly in 3D modeling and mapping applications.
Innovation Solution
The method involves identifying a higher-confidence reference object within the images, such as a stationary and easily recognizable object, and using it to improve the consistency and accuracy of camera pose data by clustering images that depict the same object and applying geolocation techniques like 3D reconstruction or geographic surveying.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing geographic positioning techniques are used to determine object locations from photographic imagery, then the process can be automated, but the accuracy of the determined locations is poor
Solution Approach 1:
The patent introduces a reference object as an intermediary element between the camera pose estimation process and the final location determination. By selecting a stationary, easily recognizable reference object visible in multiple images, the system creates a stable reference frame that mediates the alignment and consistency of pose data across different images, thereby improving both location accuracy and pose consistency without requiring manual intervention in each image
Solution Approach 2:
The patent changes the selection criteria for reference objects from arbitrary or random selection to specific parameter-based selection. By requiring reference objects to be stationary, easily recognizable, and visible in multiple images, the system transforms the pose optimization problem into a more constrained and solvable problem, improving measurement precision through parameter-based object selection
2Measurement precision
If multiple images are processed to improve location accuracy, then more data is available for analysis, but the complexity of processing and aligning the images increases
Solution Approach 1:
The patent extracts and isolates the reference object from the complex scene in multiple images. By focusing computational effort on detecting and tracking a single stationary reference object rather than processing all features in all images, the system reduces processing complexity while still utilizing multiple images to improve location accuracy through the stable reference provided by the extracted object
3Measurement precision
If camera pose data is corrected using reference objects, then the accuracy of pose data improves, but additional processing steps are required
Solution Approach 1:
The patent performs preliminary selection and identification of reference objects before the main pose optimization process. By pre-identifying stationary, easily recognizable objects that will serve as reference points, the system prepares the data in advance, making the subsequent pose correction more efficient and accurate without requiring extensive processing during the main optimization phase
Data Source
AI summary
Camera pose optimization, which includes determining the position and orientation of a camera in three-dimensional space at different times, is improved by detecting a higher-confidence reference object in the photographs captured by the camera and using the object to increase consistency and accuracy of pose data. Higher-confidence reference objects include objects that are stationary, fixed, easily recognized, and relatively large. In one embodiment, street level photographs of a geographic area are collected by a vehicle with a camera. The captured images are geo-coded using GPS data, which may be inaccurate. The vehicle drives in a loop and captures the same reference object multiple times from the substantially same position. The trajectory of the vehicle is then closed by aligning the points of multiple images where the trajectory crosses itself. This creates an additional constraint on the pose data, which in turn improves the data's consistency and accuracy.