Depth cameras calibration for obstructive sleep apnea diagnostic support

Autor: Libor Hargas, Jakub Bajzik, Silvia Janisova, Jozef Volak, Dusan Koniar
Rok vydání: 2020
Předmět:
Zdroj: 2020 ELEKTRO.
DOI: 10.1109/elektro49696.2020.9130320
Popis: In our research, we use depth cameras to obtain precise 3D models of patients’ heads and faces. Visual system based on these cameras is basic element of developed screening tool for obstructive sleep apnea. Decreasing of scanning time plays key role mainly for pediatric patients and suppresses generation of motion artifacts in 3D models. Used Time of Flight sensors are relatively non-expensive solution, especially in our case (parallel usage of several cameras). To obtain accurate models, the step of camera calibration is very important. In this article, we bring detailed look on geometrical distortion (radial and tangential) of used cameras and also methods for spatial calibration of system composed from three cameras. As software tool, we used Camera Calibrator implemented in MATLAB. As result of mentioned research, we found suitable multi-camera calibration method and set of parameters related to our visual system (given topology and geometrical properties of scanning cabin). Calibration decreased average distance error (measured using Hausdorff distances) from ca. 3 mm before calibration to ca. 2 mm after calibration and made easier the registration of partial models to the resulting one. Designed visual system based on depth cameras with proper calibration and interference suppression seems to have big potential for obstructive sleep apnea screening not only in pediatrics.
Databáze: OpenAIRE