3D Shape Detection Using Spatially Modulated Grid Points
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting the three-dimensional shape of objects suffer from ambiguities and measurement errors, particularly in selecting the correct epipolar lines for accurate triangulation, leading to suboptimal quality in 3D object recognition.
Innovation Solution
The method employs an illumination pattern with spatially modulated grid points, using an acute angle between the illumination and camera axes for triangulation, and selects the correct epipolar line by filtering distances and considering neighborhood information to ensure continuity conditions, thereby reducing ambiguities and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a structured illumination pattern with grid points is used for triangulation-based 3D detection, then distance information can be calculated, but ambiguities arise in selecting the correct epipolar lines leading to measurement errors
Solution Approach 1:
The patent applies asymmetry by introducing spatial modulation that creates asymmetric patterns around each grid point. Specifically, each grid point is assigned a unique asymmetric spatial signature through modulation in orthogonal directions, allowing unambiguous identification of the correct epipolar line even when multiple lines appear adjacent to an image grid point. This asymmetric encoding eliminates the ambiguity problem in triangulation-based 3D detection.
2Adaptability or versatility
If multiple epipolar lines are considered adjacent to an image grid point, then more potential matches are available, but selecting the correct line becomes more difficult and error-prone
Solution Approach 1:
The patent uses spatial modulation to create unique 'signatures' for each grid point, analogous to color changes. Each grid point is modulated with a distinct spatial pattern that encodes its identity. When evaluating adjacent epipolar lines, the system checks which line's associated grid point matches the observed image grid point's spatial signature. This approach makes correct line identification straightforward despite multiple candidates being geometrically adjacent.
3Reliability
If the illumination pattern contains a two-dimensional code with code elements modulated in two different spatial directions, then ambiguities are eliminated, but the device complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing the spatial signatures of all grid points in the illumination pattern. During 3D detection, the system only needs to compare the observed image grid point's position against these pre-computed signatures to identify the correct epipolar line. This preliminary preparation eliminates ambiguities during the actual measurement process without requiring complex real-time computations or additional hardware.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enhances the accuracy and reliability of 3D object recognition by unambiguously identifying epipolar lines and their associated object lattice points, resulting in improved data quality and reduced errors in determining the three-dimensional shape of objects.
Implementation Method 1
Optical radiation is projected onto a scene in the form of points and stripes. Reflected radiation is recorded via a camera
Implementation Method 2
A distance is determined using a triangulation method. This is required to use triangulation measurement principles
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a method and a device for capturing the three-dimensional shape of an object (3) illuminated by a lighting device (1) with a lighting pattern having a plurality of grid points (9), wherein the grid points (9) exhibit spatial modulation. Image grid points (C) of the illuminated object (3) are present in the image plane of a camera. It is provided that each image grid point (C) is assigned at least one spatial coordinate of the corresponding grid point (9) on the object (3). This assignment includes determining exactly one epipolar line (15) adjacent to the image grid point from a plurality of adjacent epipolar lines.Determining exactly one epipolar line (15) adjacent to the image grid point comprises determining at least one distance (δC[i, j], δ[i, j],) of the image grid point (C) or the projection (PM) of the image grid point (C) onto the respective epipolar line (15) to at least one defined point (Po, PINF) of the epipolar line (15) for a plurality of adjacent epipolar lines (15) and selecting one of these adjacent epipolar lines (15) taking into account the determined distances (δC[i, j], δ[i, j],), wherein the determined distances (δC[i, j], δ[i, j],) are filtered to exclude such epipolar lines (15) where the determined distances are below and/or above defined limit values (G1) or mean values and/or outside defined intervals (G2, G3).