Abstract: Terahertz radar features high frequency, large bandwidth, and other characteristics, thus helpful to improve the micro-motion target detection accuracy of radar systems, and feasible to extract the feature of micro-motion target. Combing with the typical micro-motion feature model, we propose a curve extraction algorithm based on the edge detection technique for the extracting the spectrum curves and further obtaining the parameters of micro-motion targets. Real data show that the proposed technique can effectively extract the feature parameters of the micro-motion target with multiple scattering centers. This research provides effective approach for further micro-motion target signal processing and target recognition.