Camera Pose Estimation Using Raw Lens Distortion Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing pose estimation methods struggle to accurately determine the dynamic translation and rotation of a camera viewpoint due to geometric distortions caused by the lens, which affect the precision of applications like SLAM, augmented reality, and robot navigation.
Innovation Solution
The method utilizes geometric distortions in raw images captured by dual or more-pixel sensors to estimate pose by analyzing phase differences between photodiodes, allowing for pose estimation at six degrees of freedom, including 3D translation and rotation, without correcting for lens distortions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If geometric distortion correction is applied to raw images, then image quality is improved, but pose estimation accuracy deteriorates due to loss of distortion information
Solution Approach 1:
The patent converts the harmful geometric distortion caused by lens into a beneficial signal for pose estimation. By analyzing the distortion characteristics in raw images before correction, the system extracts pose information that would otherwise be lost, achieving accurate pose estimation without requiring distortion correction processing.
Solution Approach 2:
The patent performs pose estimation using raw images before geometric distortion correction is applied. By extracting pose information from the undistorted image data in advance, the system avoids the need for subsequent correction processing and maintains measurement precision while simplifying the overall processing pipeline.
2Measurement precision
If traditional pose estimation methods are used, then processing pipeline is simple, but measurement precision deteriorates due to lens distortion effects
Solution Approach 1:
The patent transforms the lens distortion from a harmful factor reducing measurement precision into a useful signal for pose estimation. By leveraging the distortion characteristics in raw images, the system achieves high-precision pose estimation without adding complex correction algorithms to the processing pipeline.
Solution Approach 2:
The system uses the raw image data itself, including its inherent geometric distortions, as the source of pose information. This self-service approach eliminates the need for separate distortion correction steps, maintaining measurement precision while keeping the processing pipeline simple and efficient.
3Measurement precision
If dual or more-pixel sensors are used, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent makes the dual or more-pixel sensor serve multiple functions: it captures image data for geometric distortion analysis and simultaneously provides the phase difference information needed for pose estimation. This multi-functionality achieves high measurement precision while avoiding the need for separate sensors or complex processing systems.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enhances the accuracy of pose estimation by leveraging geometric distortions in raw images, improving precision in applications such as SLAM and augmented reality, and enabling efficient pose estimation using a single camera with reduced data processing time.
Implementation Method 1
dual or more-pixel sensors to estimate pose by analyzing phase differences between photodiodes
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
A pose estimation method and apparatus is disclosed. The pose estimation method includes acquiring a raw image before a geometric correction from an image sensor, determining a feature point in the raw image, and estimating a pose based on the feature point.