Abstract: In order to realize geomagnetism aided navigation (GAN) in the regions with insignificant geomagnetic characteristic, inspired by the pattern recognition idea, the geomagnetic matching (GM) problem was transformed into statistical pattern recognition and was analyzed from the aspect of pattern recognition. A matching algorithm based on support vector machine (SVM) was proposed. The selection of training sample for SVM, the relation between the parameter of the SVM and the geomagnetic roughness were studied. Then a SVM was constructed for GM. Simulation results show that the proposed method can greatly improve the efficiency and achieve better performance than ordinary correlative matching algorithm. The accuracy of matching method is less than the unit cell size of the geomagnetic map.