Using a 94-GHz homodyne interferometer employing a highly-directional quasi-optical lens antenna aimed at a human subject's chest, we can measure chest wall displacement from up to 10m away and through common clothing. Within the chest displacement signal are motions due to cardiac activity, respiration, and gross body movement. Our goal is to find the heart rate of the subject being monitored, which implies isolation of the minute movements due to cardiac activity from the much larger movements due to respiration and body movement. To accomplish this, we first find a subset of the true heartbeat temporal locations (called confident" heartbeats) in the displacement signal using a multi-resolution wavelet approach, utilizing Symlet wavelets. Although the chest displacement due to cardiac activity is orders of magnitude smaller than that due to respiration and body movement, wavelets find those heartbeat locations due to several useful properties, such as shape matching, high-pass filtering, and vanishing moments. Using the assumption that the confident" heartbeats are randomly selected from the set of all heartbeats, we are able to find the maximum a posteriori statistics of an inverse Gaussian probability distribution modeling the inter-heartbeat times. We then analyze the confident" heartbeats and decide which heartbeats are probabilistically correct and which are not, based on the inverse Gaussian distribution we calculated earlier. The union of the confident" set, after pruning, and the interpolated set forms a very close approximation to the true heartbeat temporal location set, and thus allows us to accurately calculate a heart rate.