This thesis project focuses on biometric recognition based on human gait analysis. During an experimental measurement campaign, radar signals were acquired from moving subjects. In parallel with the radar measurements, data were recorded from the Inertial Measurement Unit (IMU) of a smartphone held by the subject. The objective of this thesis is to investigate the relationship between radar and IMU signals and to assess the possibility of determining the consistency between the two sensing modalities. To this end, different neural network architectures are developed and compared to distinguish compatible signal pairs, belonging to the same acquisition, from non-compatible pairs, evaluating their performance in Match/No-Match classification. Finally, the model is analyzed in scenarios where the radar and IMU signals belong to different subjects, studying its behavior in terms of False Alarm and Miss Detection probabilities under different speed and pose conditions.
Il progetto di tesi si inserisce nell'ambito del riconoscimento biometrico basato sulla cinematica del movimento umano (gait analysis). Durante una campagna sperimentale sono stati acquisiti segnali radar relativi a soggetti in movimento e, parallelamente, le tracce dell'unità inerziale (IMU) di uno smartphone impugnato dal soggetto. L'obiettivo della tesi è studiare la relazione tra i segnali radar e IMU e valutare la possibilità di riconoscere la coerenza tra le due modalità di acquisizione. A tale scopo vengono sviluppate e confrontate diverse architetture di rete neurale per distinguere coppie di segnali compatibili, appartenenti alla stessa acquisizione, da coppie non compatibili, valutandone le prestazioni nella classificazione Match/No-Match. Infine, il modello viene analizzato in scenari in cui i segnali radar e IMU appartengono a soggetti differenti, studiandone il comportamento attraverso le probabilità di False Alarm e Miss Detection nelle diverse condizioni di velocità e posa.
Analisi comparativa e sintesi di segnali biometrici: identificazione dell'utente tramite sensori radar e IMU mediante tecniche di Machine Learning
FILIPPIN, ALESSANDRO
2025/2026
Abstract
This thesis project focuses on biometric recognition based on human gait analysis. During an experimental measurement campaign, radar signals were acquired from moving subjects. In parallel with the radar measurements, data were recorded from the Inertial Measurement Unit (IMU) of a smartphone held by the subject. The objective of this thesis is to investigate the relationship between radar and IMU signals and to assess the possibility of determining the consistency between the two sensing modalities. To this end, different neural network architectures are developed and compared to distinguish compatible signal pairs, belonging to the same acquisition, from non-compatible pairs, evaluating their performance in Match/No-Match classification. Finally, the model is analyzed in scenarios where the radar and IMU signals belong to different subjects, studying its behavior in terms of False Alarm and Miss Detection probabilities under different speed and pose conditions.| File | Dimensione | Formato | |
|---|---|---|---|
|
Filippin_Alessandro.pdf
accesso aperto
Dimensione
4.28 MB
Formato
Adobe PDF
|
4.28 MB | Adobe PDF | Visualizza/Apri |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/114207