How it works?

Morphological operations such as erosion.

以下内容由这个数据库的readme.txt文件翻译而来。

简介

dPPG PFT 数据集由布里斯托大学计算机科学系视觉信息实验室发布，包含 35 名受试者在进行用力肺活量（FVC）和缓慢肺活量（SVC）肺功能测试时的体表深度信息。数据集共计 300 个 PFT 序列，并包含每个序列的肺量计（spirometer）数据作为验证基准。

• 适用于：深度视觉、呼吸监测、医学信号处理等研究
• 采集方式：双 Kinect 深度相机（前后各一），同步采集
• 公开时间：2018 年 2 月
• 主要联系人：Vahid Soleimani
• 数据集主页：A Dataset for Depth-Based Whole Body Photoplethysmography in Remote Pulmonary Function Testing

• dataset: https://data.bris.ac.uk/data/dataset/1tqzx39mzkw832msuvy3obktqi

What background they describe?

• Both tests start with a few cycles of normal breathing, called tidal volume, followed by a maximal inhalation–exhalation, called main effort.
• Spirometer: Difficult for anyone to use. Recalibrated. Expensive.
• when a single depth sensor is used, decoupling trunk motion and chest-surface respiratory motion would be potentially impossible.

How they handle the data?

Short video segments with mild infant motion or those including caregiver shadow movement are classified as optimal. In contrast, videos with more significant movements, such as upper or lower limb movement (i.e., permissible motion artifacts), the presence of a parent or nurse’s hand that does not block the respiratory area, or mild crying that does not cause high motion artifacts, are considered suboptimal.

What is the application?

Home-based breathing exercises for chronic obstructive pulmonary disease (COPD) patients.

Breathing mode classification: chest and belly.

What is the novelty?

• Transferred the SOTA RGB segmentation model to depth.
• Calibration-free lung volume measurement, no clothing, and no involuntary body motion.
• Real-time (12th I5 and 4060 Ti, 3fps) (update result every 3 seconds).

What is Kalman Filter?

A recursive way to compute conditional expectation.

(The KF was discussed in page 47-48 of the paper.)

Kalman filter: This technique has been used by several studies as a sensor fusion method. Thus, it is not a method to extract breathing parameters but to fuse measurements from different sensors. When multiple respiration sensors are available, the measurements they provide are not exactly the same. Furthermore, measurements always contain noise. The Kalman filter is used to provide a final value based on the measurements of the different sensors, the model of variation of the breathing parameter, the noise model of the sensors, and the variation model [270]. Figure 33 shows an overview of the Kalman filter algorithm adapted to the fusion of breathing sensors