Computer Vision Signal Recognition in Spectrum Analyzers

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing real-time spectrum analyzers require user expertise to recognize signals, which is time-consuming and inconvenient, especially for inexperienced users, as they rely on visual identification of RF signals.

Innovation Solution

A test and measurement instrument that uses computer vision techniques, such as template matching and image processing, to automatically recognize signals by converting digital data into images and comparing them to reference images, allowing for automated signal identification without user intervention.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If user manually identifies signals based on visual appearance, then signal recognition accuracy can be achieved, but time consumption increases and requires user expertise

Engineering Contradiction:
Improvesignal recognition accuracyVSAvoidtime consumption
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent replaces the manual visual inspection process with an automated computer vision system. The spectrum display is converted into an image format, and image processing algorithms automatically analyze signal characteristics, replacing the need for human users to visually examine and identify signals manually.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system enables self-service by automatically performing signal identification without requiring user intervention. The computer vision algorithm independently analyzes the spectrum image, detects signal patterns, and identifies signal types, making the instrument autonomous in its signal recognition function.

Inventive Principle:
Principle #25Self-service

2Ease of operation

If user manually identifies signals, then signal recognition is possible, but ease of operation deteriorates especially for inexperienced users

Engineering Contradiction:
Improveease of signal identificationVSAvoiduser expertise requirement
Core Design Contradiction:
Ease of operationVSAdaptability or versatility

Solution Approach 1:

The patent substitutes the human expert's visual analysis capability with an automated image processing system. The computer vision algorithm replicates and enhances the pattern recognition abilities of experienced users, making advanced signal identification accessible to users regardless of their expertise level.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system performs preliminary analysis by automatically processing the spectrum image and identifying signal characteristics before the user needs to interpret them. This pre-processing of visual information into structured signal identification results simplifies the user's task and reduces the need for specialized knowledge.

Inventive Principle:
Principle #10Preliminary action

3Productivity

If automated signal recognition is implemented using computer vision, then productivity increases and time is reduced, but device complexity increases

Engineering Contradiction:
Improvesignal recognition speedVSAvoidsystem complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent introduces an intermediary layer by converting the spectrum data into image format. This intermediate representation allows the use of mature image processing and computer vision techniques to analyze signal patterns, bridging the gap between spectral data and automated recognition without requiring complex direct analysis algorithms.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The system creates a visual copy of the spectrum display as an image, which can then be processed using standard image recognition algorithms. This copying approach leverages existing computer vision technologies rather than developing specialized signal analysis algorithms from scratch, managing complexity while achieving automated recognition.

Inventive Principle:
Principle #26Copying

Data Source

PatentEP2386867B1Signal recognition and triggering using computer vision techniques
Publication Date: 2020.04.22 TEKTRONIX INC
  • EP2386867B1 patent drawingFigure 1~2
  • EP2386867B1 patent drawingFigure 3~4
  • EP2386867B1 patent drawingFigure 5

AI summary

A test and measurement instrument processes digital data that represents an input signal to produce a target image, and then uses a computer vision technique to recognize a signal depicted within the target image. In some embodiments, the location of the signal within the target image is identified on a display device. In other embodiments, the location of the signal within the target image is used to perform a measurement. In other embodiments, when the signal is recognized, a trigger signal is generated that causes digital data that represents the input signal to be stored in a memory.