There are numerous respiratory diseases in the United States. Traditionally, breath analysis requires contact methods which are uncomfortable, immobile, and undesirable for sustained monitoring. There is a strong need for breath measurement in a contact-free manner.

This dissertation studies the human breathing rate and waveform through thermal imaging. First, we introduced a novel Fourier analysis method to compute the signal frequency and analyze the waveform. Second, we improved the previous methodology, and applied wavelet analysis to analyze both these breathing descriptors.

We performed experiments on both laboratory and clinical subjects. We juxtaposed the computed breathing rate from thermal imaging to two traditional contact devices, the abdominal transducer belt and the thermistor. Our thermal breathing signals have high correlation with the temperature signals captured by the thermistor. In clinical sleep studies, we successfully detected sleep apnea through thermal breathing signals.

We demonstrated the feasibility of the novel thermal approach for breathing computation. Our work opens the way for unobstructive and sustained breathing monitoring in clinic.