OptFus
OptFus: Optical Sensor Fusion For The Classification of Multi-source Data
OptFus is an optical multi-sensor data fusion technique. It represents the dataset captured using different sensors with fused features. The main assumption is the datasets can be represented using a few fused common (fused) features in different subspaces. Tuning parameters control the contributions of sensors. The total variation regularizer promotes piecewise smoothness on the fused features. The subspaces learn throughout the optimization process. The estimated fused features can be used for further classification, and they considerably improve the classification results.
Citation:
OptFus: Optical Sensor Fusion For The Classification of Multi-source Data: Application to Mineralogical Mapping
doi: 10.1109/LGRS.2021.3132701
