3D Position Estimation via Multi-View Probability Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies face challenges in accurately estimating the three-dimensional position of a targeted point, especially when the point is hidden or the depth value accuracy is low, leading to potential erroneous identifications.
Innovation Solution
An information processing apparatus and method that estimate the three-dimensional position of a targeted point by identifying probability maps in multiple projection directions based on an input image, allowing for accurate positioning without relying on depth values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If depth image-based targeted point estimation is used, then three-dimensional position can be obtained, but accuracy deteriorates when targeted point is hidden or depth value accuracy is low
Solution Approach 1:
The patent transforms the problem from three-dimensional space to two-dimensional projection planes. By creating probability maps on multiple projection planes (front view, side view, top view) and identifying targeted points in two dimensions, the system avoids the reliability issues of direct three-dimensional depth-based estimation while maintaining positional accuracy through multi-planar verification.
2Measurement precision
If multiple projection directions are used to improve accuracy, then three-dimensional position estimation improves, but processing complexity increases
Solution Approach 1:
The patent divides the three-dimensional estimation problem into separate two-dimensional estimation tasks on different projection planes. Each probability map is generated and processed independently on its own projection plane, allowing parallel processing and reducing the complexity of simultaneous three-dimensional analysis while maintaining overall accuracy through integration of results from multiple views.
Data Source
AI summary
There is provided an information processing apparatus, an information processing method, and a recording medium that can estimate a three-dimensional position of a targeted point more accurately. The information processing apparatus includes: a targeted point estimation unit configured to identify probability maps indicating existence probabilities of a targeted point in a plurality of projection directions, on the basis of an input image; and a synthesis unit configured to identify a three-dimensional position of the targeted point on the basis of the probability maps in the plurality of projection directions.