A geomagnetic reference map reconstruction method based on sparse representation and dictionary learning

Geomagnetic matching navigation plays an important role in the field of navigation guidance. The construction accuracy of geomagnetic reference map determines the effectiveness of geomagnetic matching navigation. Aimed at the problem that the existing geomagnetic reference map construction accuracy is difficult to meet the needs of practical applications, a high-precision geomagnetic reference map construction method based on sparse representation and dictionary learning is proposed. First, the sparse dictionary is initialized using Rectangular Harmonic Analysis (RHA). Then, K-SVD is used to train the sparse dictionaries. Finally, the feature that the low-resolution and high-resolution reference maps have the same sparse coefficients is used to reconstruct the high-resolution geomagnetic reference maps. Experimental results show that the proposed method has higher construction accuracy for geomagnetic reference maps, lower requirements for training datasets, and better robustness to noise. Compared with the PSO-Kriging interpolation method, with a magnification factor of 4, the Peak Signal to Noise Ratio (PSNR) value is increased from 26.31 dB to 26.73 dB; the Structural Similarity Index (SSIM) is increased from 0.498 to 0.524; the Root Mean Square Error (RMSE) is decreased from 14.96 nT to 13.78 nT. © 2021, Editorial Board of JBUAA. All right reserved.