Object Localization via Parallel Projection Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object localization methods face challenges in achieving accuracy while minimizing calculation complexity, particularly in environments with non-linear distortions and varying camera settings.
Innovation Solution
A method utilizing a parallel projection model that projects objects onto virtual and actual camera planes, compensating for differences in localization between sensors and accounting for pan factors, allowing for accurate object localization with reduced complexity through iterative refinement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional localization algorithms (DLT, calibration objects) are used, then localization accuracy can be achieved, but calculation complexity and system setup complexity increase
Solution Approach 1:
The patent extracts and eliminates the need for complex calibration objects and DLT matrix calculations by using a simplified projection model. Instead of requiring calibration patterns and complex mathematical transformations, the system directly projects 3D world coordinates to 2D image coordinates using a simplified model that removes unnecessary calibration steps while maintaining localization accuracy.
Solution Approach 2:
The patent changes the mathematical parameters of the projection model from complex DLT transformations to a simplified projection equation. By modifying the projection parameters to use direct coordinate transformation without requiring 11-parameter calibration matrices, the system reduces calculation complexity while preserving measurement precision.
2Measurement precision
If calibration objects and DLT transformations are used, then accurate localization is achieved, but the system becomes more complex and requires repeated calibration when camera settings change
Solution Approach 1:
The patent removes the requirement for physical calibration objects from the system. By extracting the calibration step entirely and replacing it with a computational projection model, the system eliminates the need for operators to place and recognize calibration patterns, making the system easier to operate while maintaining accuracy.
Solution Approach 2:
The patent creates a virtual copy of the calibration process through computational projection. Instead of requiring physical calibration objects, the system uses mathematical projection to simulate the calibration relationship between camera coordinates and world coordinates, making the system easier to operate while preserving measurement precision.
3Device complexity
If pinhole camera model with single correspondence is used, then calculation is simplified, but localization accuracy is limited
Solution Approach 1:
The patent transitions from 2D image plane correspondence to 3D world coordinate projection by introducing depth information through the projection model. This dimensional enhancement allows the system to maintain the simplicity of the pinhole model while improving localization accuracy by considering the third dimension in the projection relationship.
Solution Approach 2:
The patent modifies the projection model parameters to include depth and spatial relationship information while maintaining the mathematical simplicity of the pinhole camera model. By changing the parameters to incorporate 3D projection relationships rather than just 2D correspondence, the system achieves improved accuracy without increasing model complexity.
4Measurement precision
If multiple cameras with information exchange are used, then localization accuracy is improved, but system complexity and computational load increase
Solution Approach 1:
The patent merges the projection calculations of multiple cameras into a unified projection model. By combining the projection relationships from multiple cameras into a single coordinated system, the patent achieves improved localization accuracy through multi-camera data while avoiding the complexity of separate information exchange and synchronization protocols.
Data Source
AI summary
There is provided a method of localizing an object comprising projecting an object located on an object plane and a reference point corresponding thereto on a virtual viewable plane and an actual camera plane; estimating coordinates of the reference point; and prescribing a relationship between a location of the object and the coordinates of the reference point.