Log-Polar Range Image Coarse Registration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for coarse registration of three-dimensional shape data from range images face challenges in accurately determining the positional relation among data in inconsistent coordinate systems, often requiring labor-intensive manual processes or reliance on external devices, and struggle with stability and robustness, especially in the presence of noise or occlusion.
Innovation Solution
A method utilizing a local Log-Polar range image in a log-polar coordinate system with a tangential plane, normalizing angular ambiguity through Fourier series expansion, and searching for corresponding points by nearest neighbor in a characteristic image space to determine the positional relation, enabling efficient coarse registration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual coarse registration is performed using GUI, then registration can be achieved, but labor of work increases significantly when the number of data increases
Solution Approach 1:
The system performs automatic coarse registration by computing local log-polar range images and matching them without requiring manual operator intervention. The computer automatically calculates corresponding points between range images and determines positional relations, eliminating the need for manual GUI operations while maintaining registration accuracy.
Solution Approach 2:
The manual interactive process is replaced by an automated computational system that uses log-polar coordinate transformations and image matching algorithms. The mechanical/manual operation of dragging and dropping images in GUI is substituted with automatic feature extraction and correspondence calculation based on local log-polar range image analysis.
2Measurement precision
If external devices such as GPS or turning tables are used for coarse registration, then positional relation can be obtained, but device complexity and cost increase
Solution Approach 1:
The patent extracts and utilizes only the essential geometric information directly from the range images themselves, without requiring external positioning devices. By computing local log-polar range images from the image data alone, the system eliminates the need for GPS receivers, turning tables, or other external hardware while still achieving accurate positional relation determination.
Solution Approach 2:
The local log-polar range image serves as an intermediary representation that captures the essential geometric and positional information needed for registration. This intermediate form enables direct comparison and matching between images without requiring external device measurements, acting as a mediator that translates raw image data into registration-ready features.
3Ease of operation
If curvature-based methods are used for coarse registration, then registration can be performed, but stability decreases in the presence of noise or occlusion
Solution Approach 1:
The patent transforms the range image data into log-polar coordinates, changing the dimensional representation from Cartesian to log-polar space. This transformation provides rotation invariance and enhances the stability of feature matching by representing angular and radial information in a manner that is less sensitive to noise and occlusion, while maintaining automatic registration capability.
Data Source
AI summary
As a feature, a local range image described in a log-polar coordinate system with a tangential plane set as an image plane is used. In a created image, ambiguity concerning an angular axis around the normal is normalized as a power spectrum using Fourier series expansion and changed to an amount invariable with respect to rotation. The power spectrum is dimensionally compressed by expanding the power spectrum in a peculiar space using a peculiar vector. Corresponding points are searched by nearest neighbor in a dimensionally compressed space to calculate a correspondence relation among the points. Wrong correspondence is removed by verification to determine a positional relation among range images. A reliable correspondence relation is narrowed down by verification by cross-correlation and a RANSAC to create a tree structure representing a link relation among the range images. A shape mode is created by applying a simultaneous registration method to plural range images of the tree structure using a result of this registration as an initial value.