In order to solve the problem of magnetic anomaly detection (MAD), many scholars have proposed the corresponding detection methods. These methods can only detect the presence or absence of the target, while obtaining anomaly signal itself can achieve the goal of locating and identifying the target. In this paper, the magnetic anomaly signal extraction model based on airborne platform and the frequencydomain correlation extraction algorithm based on convolution blind source separation (CBSS) are proposed. The extraction of weak magnetic anomaly signal is implemented. Simulation results demonstrate the feasibility and effectiveness of the proposed method.