Multi-Camera Object Identification Using Expected Location Tracking

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Computer vision systems with multiple cameras face high computational burdens in object identification due to the need for visual feature matching across large search spaces and multiple target objects, which is inefficient and resource-intensive.

Innovation Solution

A multi-camera system that uses visual feature matching in one camera's field of view to identify objects and then tracks these objects across different camera views using expected locations, reducing the need for repeated visual feature matching by sharing object identities based on spatial and temporal associations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If visual feature matching is performed across multiple camera fields of view to identify objects, then object identification accuracy is improved, but computational burden increases significantly

Engineering Contradiction:
Improveobject identification accuracyVSAvoidcomputational burden
Core Design Contradiction:
Measurement precisionVSPower

Solution Approach 1:

The system performs visual feature matching only in the first camera's field of view to identify objects, then uses the identified object's expected location to guide searching in subsequent camera fields of view. This preliminary identification action eliminates the need to perform computationally intensive visual feature matching in every camera, thereby reducing overall computational burden while maintaining identification accuracy

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent extracts only the necessary information (object identity and expected location) from the first camera's identification result, and uses this extracted information to guide the search in other cameras. This selective extraction approach avoids redundant processing and reduces computational load compared to performing full visual feature matching across all cameras

Inventive Principle:
Principle #2Taking out (Extraction)

2Reliability

If visual feature matching is performed in every camera's field of view, then object tracking reliability is improved, but processing time increases

Engineering Contradiction:
Improveobject tracking reliabilityVSAvoidprocessing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system performs preliminary object identification in the first camera before processing other cameras. This preliminary action establishes the object's identity and expected location, allowing subsequent cameras to use this information to efficiently locate and track the same object without repeating the full identification process, thereby reducing processing time while maintaining tracking reliability

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

Instead of performing complete visual feature matching in every camera, the system performs partial matching using the expected location information from the first camera. This partial action approach is sufficient for tracking purposes and significantly reduces processing time compared to exhaustive matching in all cameras

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP4290472A1Object identification
Publication Date: 2023.12.13 NOKIA TECHNOLOGIES OY
  • EP4290472A1 patent drawingFigure 1~2B
  • EP4290472A1 patent drawingFigure 3~4
  • EP4290472A1 patent drawingFigure 5

AI summary

A system comprising: multiple cameras including at least a first camera having a first field of view and a second camera having a second field of view, wherein the second camera is different to the first camera and the second field of view is different to the first field of view; and identification means for identifying an object captured by one or more of the multiple cameras, wherein the identification means comprises means for: using visual feature matching for a detected object in the first field of view of the first camera to identify the detected object in the first field of view of the first camera as a first object; and using an expected location of the first object in the second field of view of the second camera to identify a detected object in the second field of view as the first object.