3D Object Localization with Virtual Cameras for Distance Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for estimating three-dimensional positions of objects suffer from reduced accuracy, particularly in determining the distance to an object, and existing gaze estimation systems lack precision in object positioning.
Innovation Solution
A method involving the use of a physical camera to capture an image, detect target objects, generate virtual cameras based on object coordinates, and apply a trained model to estimate three-dimensional positions using probability distributions, constrained by predetermined parameters and calibration settings to enhance accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods are used to estimate three-dimensional positions from images, then the process is simple, but the accuracy of distance estimation is reduced
Solution Approach 1:
The patent introduces virtual cameras as intermediary computational models between the physical camera and the target object. These virtual cameras simulate different viewing angles and positions, allowing the system to estimate three-dimensional position with higher accuracy by combining multiple virtual perspectives without requiring multiple physical cameras
Solution Approach 2:
The system changes parameters by creating virtual cameras with varying positions, orientations, and focal lengths. By adjusting these camera parameters computationally, the system can estimate depth and distance more accurately from a single physical image, transforming a 2D image problem into a multi-perspective 3D estimation problem
2Measurement precision
If multiple physical cameras are used to improve three-dimensional position accuracy, then measurement precision improves, but device complexity and cost increase
Solution Approach 1:
The patent creates virtual copies of cameras (virtual cameras) that simulate multiple viewing perspectives. Instead of deploying multiple physical cameras, the system generates computational copies with different positions and orientations, processing a single physical image through multiple virtual camera models to achieve three-dimensional position estimation with the accuracy of multi-camera systems
Solution Approach 2:
The system replaces the mechanical approach of using multiple physical cameras with a computational approach using virtual cameras. The physical camera captures a single image, and then computational algorithms generate multiple virtual views and estimate three-dimensional position, substituting mechanical complexity with software-based processing
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The disclosure relates to a method (300) of estimating a three-dimensional position (150) of at least one target object (111) performed by a computer, the method comprising obtaining a first image (130), detecting at least one target object (111), generating a set of images depicting subsets of a captured scene, and estimating a three-dimensional position of at least one target object using a probability distribution (P) indicative of a likelihood that the at least one target object is in a particular position (x, y, z) in the scene.