Moving Body Position Estimation via Dynamic Feature Point Density
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing moving body position estimating devices face challenges in increasing precision for changes in the surrounding environment, particularly in accurately estimating the position and orientation of a moving body in three-dimensional space using images from a camera mounted on the body.
Innovation Solution
The solution involves a device and method that acquire and process first and second images to estimate the direction of rotation and change in position, detect feature points within specific regions of the images, and determine corresponding points to estimate the position change of the moving body, with dynamic adjustment of feature point density based on the estimated rotation or translation direction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If feature points are detected uniformly across the entire image, then the detection coverage is comprehensive, but the processing time increases and precision for motion estimation decreases
Solution Approach 1:
The image is divided into multiple regions (first region, second region, third region) with different feature point detection densities. The first region has higher feature point density than the second region, allowing selective processing that reduces overall computation while maintaining precision where needed.
Solution Approach 2:
Different regions of the image are assigned different quality levels in terms of feature point detection density. The first region (with higher density) is used for directions with larger motion components, while the second region (with lower density) is used for directions with smaller motion components, optimizing both precision and processing speed.
2Measurement precision
If feature point density is increased to improve position estimation precision, then measurement accuracy improves, but processing complexity and time increase
Solution Approach 1:
The feature point detection density is dynamically adjusted based on the estimated motion direction. The processor estimates the rotation direction of the moving body and selectively applies high density detection in the first region for directions with larger motion components, while using lower density in the second region for directions with smaller motion components.
Solution Approach 2:
The detection parameters (feature point density) are changed according to the estimated motion characteristics. By changing the density parameter based on rotation direction estimation, the system achieves high precision where needed while reducing overall processing complexity.
3Measurement precision
If feature points are detected in all regions with high density, then position estimation precision improves, but the system becomes less adaptable to different motion scenarios
Solution Approach 1:
The system dynamically adapts feature point detection strategies based on estimated motion characteristics. By estimating rotation direction and adjusting detection density in different regions accordingly, the system maintains high adaptability to various motion scenarios while optimizing precision.
Solution Approach 2:
The system uses feedback from rotation direction estimation to adjust feature point detection parameters. The estimated motion direction informs where high-density detection is needed, creating a closed-loop system that adapts to different motion scenarios.
Data Source
AI summary
A moving body position estimating device includes an acquisition unit and a processor. The acquisition unit acquires information including first data relating to a first image and second data relating to a second image including a difference between the first image and the second image accompanying a movement of a moving body. The processor implements estimating, based on the information, a direction of a rotation accompanying the movement; detecting, based on the direction of the rotation, first feature points inside a first region inside the first image and second feature points inside a second region inside the first image; and determining first corresponding points inside a third region inside the second image and estimating a change of a position of the moving body based on each of the first feature points and each of the first corresponding points.