Low-rank tensor learning for classification of hyperspectral image with limited labeled samples.

•Propose a low-rank tensor learning (lrTL) method for HIS classification.•Obtain the small local patches by superpixel segmentation.•Obey the 3D natural of the HSI cube by third-order tensor representation.•Capture the global structure of the HSI by low-rank constraint.•Validate effectiveness of the lrTL by experimental analysis.