Fall Detection System Based on Point Cloud Enhancement Model for 24 GHz FMCW Radar

Tingxuan Liang,Ruizhi Liu,Lei Yang,Yue Lin,C.-J. Richard Shi,Hongtao Xu

DOI: https://doi.org/10.3390/s24020648

IF: 3.9

2024-01-20

Sensors

Abstract:Automatic fall detection plays a significant role in monitoring the health of senior citizens. In particular, millimeter-wave radar sensors are relevant for human pose recognition in an indoor environment due to their advantages of privacy protection, low hardware cost, and wide range of working conditions. However, low-quality point clouds from 4D radar diminish the reliability of fall detection. To improve the detection accuracy, conventional methods utilize more costly hardware. In this study, we propose a model that can provide high-quality three-dimensional point cloud images of the human body at a low cost. To improve the accuracy and effectiveness of fall detection, a system that extracts distribution features through small radar antenna arrays is developed. The proposed system achieved 99.1% and 98.9% accuracy on test datasets pertaining to new subjects and new environments, respectively.

engineering, electrical & electronic,chemistry, analytical,instruments & instrumentation

What problem does this paper attempt to address?

The paper aims to address the issue of fall detection for the elderly, particularly in indoor environments. Current fall detection methods have some shortcomings, such as wearable sensors that may be forgotten to be worn, and vision-based methods that are costly and invade privacy. In contrast, millimeter-wave radar sensors offer advantages such as privacy protection, low hardware cost, and wide operating conditions, making them an attractive choice.

However, the point cloud data obtained using 24GHz FMCW radar is of low quality, affecting the accuracy of fall detection. Traditional methods improve detection accuracy by using more expensive hardware, but this approach is not cost-effective. To solve this problem, the paper proposes a new Point Cloud Enhancement (PCE) model that can generate high-quality 3D point cloud images from low-cost radar and extract distribution features using a small radar antenna array to improve the accuracy and effectiveness of fall detection.

Specifically, the system achieves this through the following steps:
1. **Preprocessing**: Accumulate point cloud data over a period of time and expand the number of point clouds to a fixed number through zero-padding.
2. **Point Cloud Enhancement (PCE) Model**: Remove noise points and generate high-quality point clouds.
3. **Feature Extraction**: Extract human posture parameters using the generated high-quality point clouds.
4. **Classifier**: Use a lightweight classifier to classify normal events and fall events.

The system achieved an accuracy of 99.1% and 98.9% on datasets of new environments and new subjects, respectively.