Cross-Camera Person Identification Using Positional Correspondence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing monitoring camera systems struggle with identifying objects in overlapping areas without pre-installed markers, leading to difficulties in markerless object identification and increased processing load and accuracy issues.
Innovation Solution
An information processing apparatus generates a correspondence relationship between image elements of multiple cameras based on detected object positions, enabling markerless object identification by collating feature information and using positional correspondence relationships to associate and identify objects across different camera views.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If markers are installed on the floor surface for coordinate conversion, then object identification accuracy is improved, but installation complexity and cost increase
Solution Approach 1:
The patent extracts and removes the markers from the identification system. Instead of requiring physical markers on the floor, the system uses natural features of the environment and mathematical models to achieve coordinate conversion and object identification, thereby eliminating installation complexity while maintaining accuracy
Solution Approach 2:
The patent introduces an intermediary coordinate conversion model that maps positions between different camera views without requiring physical markers. The model uses camera parameters and geometric relationships as intermediaries to translate object positions across multiple camera coordinate systems
2Area of stationary object
If multiple cameras are used for tracking, then coverage area is improved, but processing load increases
Solution Approach 1:
The patent segments the processing task by assigning different cameras to monitor specific regions. Each camera processes only the objects within its field of view, and the system uses coordinate conversion to track objects across camera boundaries, thereby distributing processing load while maintaining comprehensive coverage
Solution Approach 2:
The patent processes only the necessary portions of data from each camera - specifically, only objects detected in each camera's view are processed and tracked. The system avoids processing redundant data by using coordinate conversion to determine which objects require cross-camera tracking
3Measurement precision
If coordinate conversion is performed for each frame image, then identification accuracy is improved, but processing time increases
Solution Approach 1:
The patent performs preliminary coordinate conversion setup during system initialization, where camera parameters and transformation matrices are pre-calculated and stored. During actual operation, the system uses these pre-computed parameters for rapid coordinate transformation, avoiding repeated complex calculations for each frame while maintaining accuracy
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Disclosed is an identification method of causing a computer to execute processing of generating, when a first person is detected from a first image captured by a first camera and a second person is detected from a second image captured by a second camera, relationship information obtained by associating a position in the first image from which the first person is detected with a position in the second image from which the second person is detected, and identifying, based on feature information on the first person and feature information on the second person, the first person and the second person.