Tang, Chong;
(2023)
3D Human Skeletal Motion Reconstruction Using WiFi Micro-Doppler Signature.
Doctoral thesis (Ph.D), UCL (University College London).
Abstract
With a growing interest in indoor sensing, passive WiFi-based RF sensing technology has attracted a great deal of attention in many areas. The characteristics of low cost, covertly monitoring and tracking non-cooperative targets make it have prospective applications in, for example, crime monitoring scenarios. In practice, the RF sensing system obtains micro-Doppler (µ-Doppler) signatures from the cross-correlation between signal returns and source signals. The captured information can be presented in the form of the µ-Doppler spectrogram. In this process, complex signal processing and µ-Doppler spectrogram interpretation work need to be completed. However, the main challenge is that traditional processing algorithms are inefficient to obtain motion details as expected. In recent years, deep learning algorithms have achieved great success in many fields. It can thoroughly analyse complex information and obtain models that can be applied to new inputs. Compared with traditional algorithms, deep learning is an experience-based model which better fits with complex RF sensing scenarios. Therefore, this thesis proposes MDPose, a novel deep learning-based framework for human skeletal motion reconstruction based on WiFi µ-Doppler signatures. MDPose provides an effective solution to represent human activity by reconstructing a skeleton model with 17 key points, which can assist with the interpretation of conventional RF sensing outputs in a more understandable way. MDPose is implemented over three sequential stages to address a series of challenges: First, denoising algorithms are employed to remove unwanted interference that may affect feature extraction and enhance weak Doppler measurements. Secondly, a convolutional neural network-recurrent neural network is applied to learn temporal-spatial dependency from cleaned µ-Doppler and estimate velocity information. Finally, a pose optimisation mechanism is proposed to determine the initial state of the skeleton and to limit accumulated errors. From the experimental result, MDPose has achieved promising performance and outperformed state-of-the-art RF motion tracking systems in many aspects.
| Type: | Thesis (Doctoral) |
|---|---|
| Qualification: | Ph.D |
| Title: | 3D Human Skeletal Motion Reconstruction Using WiFi Micro-Doppler Signature |
| Language: | English |
| UCL classification: | UCL UCL > Provost and Vice Provost Offices > UCL BEAMS UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science > Dept of Security and Crime Science |
| URI: | https://discovery-pp.ucl.ac.uk/id/eprint/10176089 |
Archive Staff Only
 |
View Item |
