3D Object Recognition via Geometric Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current object recognition methods are limited to two-dimensional approaches and fail to identify the third dimension of an object, making it difficult to recognize objects in digital space regardless of their orientation.
Innovation Solution
A system that uses machine learning and computer-executable instructions to create a database for three-dimensional object recognition, allowing objects to be identified and differentiated based on their geometry and structure from any orientation by simulating rotations across all axes and performing mathematical calculations to extract comparative data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If two-dimensional approaches are used for object recognition, then the recognition process is simple and successful on social media websites, but the third dimension of objects cannot be identified
Solution Approach 1:
The patent transitions from two-dimensional image analysis to three-dimensional object recognition by extracting geometric features from 3D digital models. The system processes X, Y, Z coordinates and spatial relationships to enable identification of objects in three-dimensional space, adding the missing dimensional capability while maintaining recognition accuracy.
2Adaptability or versatility
If three-dimensional object recognition is implemented, then objects can be identified from any orientation, but the system complexity increases significantly
Solution Approach 1:
The patent segments the three-dimensional object recognition problem into distinct processing stages: extracting geometric features from 3D models, normalizing spatial coordinates, identifying shape descriptors, and comparing feature sets. This segmentation breaks down the complex task into manageable components that can be processed systematically.
Solution Approach 2:
The system performs preliminary processing of three-dimensional models by pre-extracting geometric features and creating normalized representations before actual recognition occurs. Testing data is constructed in advance with known geometric properties, enabling the machine learning system to learn from pre-processed three-dimensional information rather than raw model data during operation.
3Measurement precision
If geometric features are extracted from three-dimensional models, then comprehensive object identification is enabled, but data processing requirements increase
Solution Approach 1:
The patent extracts only the essential geometric features from complete three-dimensional models, such as shape descriptors, spatial relationships, and key dimensional properties. Rather than processing entire model datasets, the system identifies and extracts specific geometric characteristics that are sufficient for recognition, reducing the volume of data that needs to be processed while maintaining recognition accuracy.
Data Source
AI summary
A computer-implantable method includes accessing a model file containing a digital 3-D object, calculating at least one dimensional measurement of the object, uniformly scaling the object in the X, Y and Z axes by a predetermined percentage, mirroring the object across the X, Y and Z axes, slicing the mirrored object at a predetermined interval by an infinite plane in both a rotated and animated state of the infinite plane, generating from the slices points at the edge of a plane object collision, and assigning to the points positive and negative values relating to all possible X, Y and Z quadrant locations.