Camera Position Correction Using Depth and Projection Images

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional camera position estimation methods struggle to accurately estimate camera positions when illumination conditions change or large errors are present in 3D points.

Innovation Solution

An information processing apparatus and method that generates a projection image and a depth image based on a query image and 3D scene information, and corrects the camera position using the depth image and correspondence relationship between the query image and projection image.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional PnP problem solving method is used to estimate camera position, then the estimation process is simple, but the accuracy deteriorates when illumination conditions change or large errors are included in 3D points

Engineering Contradiction:
Improvecamera position estimation accuracyVSAvoidstability under illumination change
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent performs preliminary actions by generating a projection image and depth image from the query image before solving the PnP problem. These pre-processed images serve as reference data that improves the accuracy of camera position estimation, especially under varying illumination conditions. The depth image provides three-dimensional information that helps maintain reliability even when illumination changes or 3D points contain errors.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces intermediary elements (projection image and depth image) that mediate between the query image and the camera position estimation process. These intermediary images act as a bridge, providing additional geometric and photometric constraints that improve measurement precision while maintaining reliability under adverse conditions.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If conventional camera position estimation is used, then computational complexity is low, but measurement precision deteriorates when errors are present in 3D points

Engineering Contradiction:
Improvecamera position accuracyVSAvoidprocessing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system performs preliminary image generation (projection image and depth image) before the main estimation task. While this adds processing steps, it significantly improves measurement precision by providing robust reference data that compensates for errors in 3D points, making the additional complexity worthwhile for achieving accurate camera position estimation.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent employs feedback mechanisms where the generated projection image and depth image are used to refine the camera position estimation. This iterative feedback process improves measurement precision by continuously adjusting the estimation based on the reference images, thereby justifying the increased processing complexity through superior accuracy.

Inventive Principle:
Principle #23Feedback

Data Source

PatentUS20250045955A1Information processing apparatus, information processing method, and computer-readable recording medium
Publication Date: 2025.02.06 NEC CORP
  • US20250045955A1 patent drawing
  • US20250045955A1 patent drawing
  • US20250045955A1 patent drawing

AI summary

An information processing apparatus includes an image generating unit that generates, based on a camera position of a query image that is obtained by shooting a scene by a camera from the camera position, and three dimensional information regarding the scene, a projection image and a depth image that correspond to shooting from the camera position, and a camera position correcting unit corrects the camera position based on the depth image and a correspondence relationship between the query image and the projection image.