In order to improve the accuracy and reliability of electronic compass heading angle measurement, a multi magnetometer electronic compass architecture is proposed, and an error compensation and fusion algorithm based on Recursive Least Squares (RLS) combined with Kalman Filter (KF) is designed using the spatial constraints of the geomagnetic field and magnetometer. Firstly, a multi magnetometer error model is constructed based on the measurement error characteristics of the geomagnetic field, and RLS is used for ellipsoidal fitting to obtain error compensation parameters. Then, KF iterative fusion is performed on isomorphic multi-sensor data based on the noise covariance matrix. Finally, simulation and actual experiments have verified the effectiveness of the multi magnetometer model and compensation fusion algorithm. The final heading angle measurement accuracy is within ±0.5°, with a standard deviation of less than 0.2°, which is about three times higher than that of a single magnetometer.