A spatial–temporal Hopfield neural network approach for super-resolution land cover mapping with multi-temporal different resolution remotely sensed images

摘要: Abstract The mixed pixel problem affects the extraction of land cover information from remotely sensed images. Super-resolution mapping (SRM) can produce maps with a finer spatial resolution than images, and reduce to some extent. Traditional SRMs solely adopt single coarse-resolution image as input. Uncertainty always exists in resultant fine-resolution maps, due lack about detailed patterns. development remote sensing technology has enabled storage great amount fine These data provide are promising reducing SRM uncertainty. This paper presents spatial–temporal Hopfield neural network (STHNN) based SRM, by employing both current previous map STHNN considers information, well temporal sub-pixel pairs distinguishing unchanged, decreased increased fractions each pixel, uses different rules labeling these sub-pixels. proposed method was tested using synthetic images class fraction errors real Landsat comparing pixel-based classification several popular methods including pixel-swapping algorithm, change method. Results show that outperforms method, algorithm model most cases. weight parameters constraints, constraints constraint have important functions performance. heterogeneity degree affect accuracy, be served guidances selecting optimal parameters.