Camera Pose Estimation Using Feature Scale Depth Ratios
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current relative pose estimation methods in computer vision, particularly for camera pose estimation, face challenges in accurately determining translation and rotation between camera poses, especially when scale information is not fully utilized, leading to instability and errors in 3D reconstruction and odometry applications.
Innovation Solution
The method extracts features from images, incorporating both location and scale information to calculate depth ratios, allowing for the selection of a reduced number of matched features for estimating camera pose, and employs robust estimation frameworks like RANSAC to refine model parameters, effectively handling errors and outliers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If classic algorithms (8-point, 7-point, 6-point, or 5-point) are used to recover pose from point correspondences, then the method is well-established and computationally efficient, but accuracy deteriorates when scale information is not utilized and intrinsic calibration is unknown
Solution Approach 1:
The patent changes the parameters used for pose estimation by incorporating scale information from detected features. Instead of relying solely on 2D point correspondences, the method uses scale ratios derived from feature detection to estimate depth ratios, thereby improving accuracy without requiring known intrinsic calibration parameters.
Solution Approach 2:
The patent introduces depth ratio as an intermediary parameter that connects scale information to 3D reconstruction. By estimating depth ratios from scale ratios of matched features, the method bridges the gap between 2D image features and 3D pose estimation, enabling accurate reconstruction without direct access to intrinsic calibration data.
2Measurement precision
If more matched features are selected for pose estimation, then estimation accuracy improves, but computational cost and processing time increase
Solution Approach 1:
The patent applies local quality by selectively using scale information from specific matched features. Rather than processing all features uniformly, the method identifies and utilizes features with reliable scale information to estimate depth ratios, thereby improving accuracy while minimizing computational overhead.
3Measurement precision
If scale information from feature detection is used to estimate depth ratios, then 3D reconstruction accuracy improves, but errors in scale estimation can propagate to pose estimation
Solution Approach 1:
The patent implements feedback by using the estimated pose and depth information to refine and verify depth ratio estimates. The method checks consistency between depth ratios derived from scale information and those implied by the reconstructed 3D geometry, allowing correction of errors and improving robustness against scale estimation inaccuracies.
Data Source
AI summary
A camera pose estimation method determines the translation and rotation between a first camera pose and a second camera pose. Features are extracted from a first image captured at the first position and a second image captured at the second position, the extracted features comprising location, scale information and a descriptor, the descriptor comprising information that allows a feature from the first image to be matched with a feature from the second image. Features are matched between the first image and the second image. The depth ratio of matched features is determined from the scale information. n matched features are selected, where at least one of the matched features is selected with both the depth ratio and location. The translation and rotation are calculated between the first camera pose and the second camera pose using the selected matched features with depth ratio derived from the scale information.