Enhancements of Non-Restrictive Visual Sensing System for Reliable Monitoring of Respiration Patterns

概要

論文の詳細を見る
We have developed Visual Sensing System that monitors respiration for 24 hours non-restrictively. The principle of this system is inter-image subtraction. The gray scale levels of pixels on a ROI (region of interest) of a current frame are subtracted from those of a previous frame to make respiration patterns in time domain. Compared with conventional restrictive methods, this method is more convenient for long respiration monitoring since it does not restrain patients' bodies. In this paper, we explain some methods to enhance the system performance. We first explain methods that improve accuracy of detecting respiration patterns by setting a threshold and an interval for inter-image subtraction most adaptively. We then explain an authenticity judgement method that judges the authenticity of respiration patterns by comparing frequency characteristics of current periodic movements with those of normal respiration using Fast Fourier Transform and by comparing gray scale histograms of the current and the previous frames. We lastly explain an emergency message and a data sending method that sends emergency messages and respiration data to host computers using a socket function when abnormality in respiration is detected. Though the algorithms are simple, we achieved a high precision in extracting respiration patterns and also satisfied the needs of patients living at distant places.
一般社団法人日本機械学会の論文
1999-09-15