Geomagnetism aided inertial navigation system for unmanned aerial vehicles was studied. An analysis was presented for the whole magnetic matching process. “One-dimension matching” concept was first used and analyzed in view of its feature space, similarity metric, search space and search strategy, upon which a framework for matching methods study was built. Through simplification of inertial computational process, one-dimension magnetic matching can be considered as an affine transformation problem with compensation. A contour constraint based iterative coarse fine two-stage matching method was proposed to get the best transformation parameters, in which global search ability and localization precision were both considered. Simulation and real experiment results show the improvements in matching precision, speed and adaptability.