Consistency Adjustment of Time-Series Land Cover Products Based on Matrix Decomposition

In recent years, the continuous increase in the number of earth observation satellites has led to the development of time-series land cover products. To obtain a land cover map with better spatiotemporal consistency and classification accuracy, this paper proposes a time series land cover map consistency adjustment method based on matrix decomposition. Matrix decomposition refers to the decomposition of a matrix into the product of several matrices, whose sizes are usually simpler than the original matrix and can represent the features of the original matrix. To verify the effectiveness of this method, Gaofen-1 satellite time series images of Shunyi District, Beijing were used for consistency adjustment. The final results show that the overall accuracy of the confusion matrix of the SVM (Support Vector Machine) classification results is 85.3952% while the overall accuracy of the confusion matrix increases to 90.3164% after consistency adjustment, and the Kappa coefficient is 0.8537. The accuracy of land cover maps and the consistency between different times of maps have been significantly improved.