Visual SLAM Structural Feature Matching From Image Edges
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current visual SLAM methods based on image point or line features suffer from low reliability due to limited information, leading to frequent false matching and high computational requirements, as they rely on extracting and matching numerous features from images.
Innovation Solution
The method employs edge information to determine structural features of components in images, including edge curves and key points, which are then matched between frames to improve localization accuracy and reduce computational load by focusing on fewer, more descriptive features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If point features or line features are used for visual SLAM, then feature extraction and matching can be performed, but the features contain limited information leading to low reliability and frequent false matching
Solution Approach 1:
The patent combines multiple edge curves and their geometric relationships into a unified structural feature representation. By merging adjacent edge curves and analyzing their spatial relationships (parallel, perpendicular, intersecting), the system creates composite features that contain richer information than individual point or line features, thereby improving matching reliability while reducing information loss.
Solution Approach 2:
The patent creates composite structural features by combining multiple edge curve elements with their geometric relationships. These composite features integrate information from multiple source elements (edge curves, key points, geometric relationships) to form a more robust and information-rich feature representation, analogous to creating composite materials with enhanced properties.
2Reliability
If a large quantity of features are extracted to compensate for low single feature reliability, then more features are available for matching, but the computational load increases significantly
Solution Approach 1:
By merging multiple edge curves into structural features, the patent reduces the total number of features that need to be processed. Instead of extracting and matching many individual point features, the system extracts fewer structural features that encapsulate information from multiple edge curves, thereby maintaining reliability while improving computational efficiency.
Solution Approach 2:
The patent segments the image processing task into edge detection, edge curve extraction, and structural feature formation stages. This segmentation allows the system to focus computational resources on extracting meaningful structural relationships rather than processing every individual feature point, improving overall computational efficiency while maintaining feature quality.
3Measurement precision
If point features and line features are combined as in PL-SLAM, then localization precision is improved, but the features still lack sufficient discriminative information to avoid false matching
Solution Approach 1:
The patent transitions from 0D point features and 1D line features to 2D structural feature representations that incorporate spatial relationships and geometric configurations. By adding this dimensional aspect that captures the arrangement and relationships between edge curves, the system achieves better discrimination capability while maintaining localization precision.
Solution Approach 2:
The patent creates composite structural features that integrate information from multiple edge curves and their geometric relationships. These composite features provide richer discriminative information compared to simple point or line features, enabling the system to distinguish between similar structures more effectively and reduce false matching.
Data Source
AI summary
Embodiments of this application disclose a method, a computer device, and a storage medium for simultaneous localization and mapping, applied to the field of information processing technologies. In the simultaneous localization and mapping method in the embodiments, edge information of a current frame image captured by an image capturing apparatus is obtained, and the edge information is divided into structural features of a plurality of first components, where each first component may correspond to a partial structure of one object in the current frame image, or correspond to at least one object. Then, a correspondence between first components in the current frame image and second components in a reference frame image may be obtained by matching the structural features of the first components with structural features of the second components in the reference frame image. Finally, the image capturing apparatus may be simultaneously localized according to the correspondence.