<p>Radar-based contactless detection of detailed cardiac activity enables convenient and accurate heart monitoring in healthcare applications. However, it requires further investigation regarding the reproducibility of stable characteristic cardiogram waveforms. In this study, we initially perform a systematic link-budget analysis for radar-based cardiogram detection to derive the performance requirements for the radar system to be used in the detection of weak cardiogram signals. Then, we propose a novel velocity cardiogram (VCG) extraction framework, which employs vector analytic demodulation (VAD)-Wiener filtering, to accurately extract cardiograms based on the signals obtained using an interferometric biomedical radar. The obtained VCG waveforms provide more stable cardiogram waveforms than Doppler cardiogram waveforms in terms of reproducibility. We conducted clinical experiments to obtain cardiac activity data from individuals with various body types and ages, covering 6123 cardiac cycles in total. The results showed that the proposed framework provides stable velocity cardiogram waveforms, which can be used to extract physiologically meaningful characteristics. In addition, the extracted RR intervals from all detected cardiac activities show a correlation of 0.987 with the ground-truth ECG, validating the high stability and reliability of the proposed cardiogram detection framework.</p>