Mobile Body Motion Calculation Using Segmented Correspondence Points
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for calculating the motion of a mobile body using images captured by a camera suffer from errors due to incorrect extraction of corresponding points and instability in parameter estimation, especially when the road surface is inclined or uneven, leading to inaccurate motion calculations.
Innovation Solution
A two-stage motion calculation method is employed, where a first motion is calculated assuming a predetermined plane, and a second motion is derived using the first motion and corresponding points, with the first motion being evaluated using partial corresponding point sets and a search technique to reduce errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a method of using images captured by a camera is employed to calculate mobile body motion, then the method can be used in principle to correctly calculate motion even when the road surface is not horizontal or the mobile body is inclined, but measurement errors such as large errors, noise, and quantization errors occur in image processing
Solution Approach 1:
The patent segments the corresponding points into multiple groups based on their spatial distribution and characteristics. By dividing the set of corresponding points into different groups and calculating motion parameters for each group separately, the method reduces the impact of local errors and noise on the overall motion calculation, thereby improving measurement precision while maintaining adaptability to inclined surfaces.
Solution Approach 2:
The patent uses more corresponding points than the minimum required (excessive action) and selects partial sets of corresponding points to perform multiple calculations. By using a larger number of corresponding points and performing multiple partial calculations, the method suppresses the influence of individual measurement errors through statistical averaging, improving the reliability of motion calculation on inclined surfaces.
2Productivity
If the eight-point method is used to calculate camera motion from corresponding points, then camera motion can be calculated using 8 or more corresponding static points, but it is difficult to extract 8 or more corresponding points with small error and the instability of the expression leads to large calculation errors
Solution Approach 1:
The patent divides the corresponding points into multiple groups and performs motion calculation for each group separately. This segmentation approach reduces the complexity of extracting and matching individual corresponding points, as each group requires fewer precise matches. The segmented calculation also stabilizes the expression by distributing the computational load, reducing the impact of extraction errors on the final result.
Solution Approach 2:
The patent uses more corresponding points than the minimum eight required by the traditional eight-point method. By using excessive corresponding points and performing multiple partial calculations with different point sets, the method compensates for extraction errors through redundancy. This approach maintains high productivity by efficiently utilizing available image data while improving precision through error suppression.
3Reliability
If corresponding points are extracted from captured images, then motion calculation can be performed, but measurement errors such as large errors, noise, and quantization errors substantially prevent elimination of errors in the calculation
Solution Approach 1:
The patent segments the corresponding points into multiple groups and calculates motion parameters for each group independently. This segmentation isolates measurement errors to specific groups, preventing a single erroneous point from affecting the entire calculation. By combining results from multiple segmented calculations, the method improves reliability while compensating for individual measurement precision limitations.
Solution Approach 2:
The patent uses more corresponding points than the minimum required and performs multiple partial calculations with different point sets. This excessive use of corresponding points creates redundancy that allows error suppression through statistical methods. The multiple partial calculations with different point combinations enable the system to identify and eliminate outliers, improving reliability despite individual measurement errors.
Data Source
AI summary
In a mobile body motion calculation apparatus (100), a corresponding point calculation section (101) calculates corresponding points between images captured by a camera 120. A first motion calculation section (102) calculates a first motion (Ma) of a mobile body using the corresponding points, assuming a predetermined plane in the images. A second motion calculation section (103) calculates a second motion (Mb) using the first motion (Ma) and the corresponding points.