Low-Rank Decomposition Model for Adaptive Identification of Similar Neighboring Pixels in Hyperspectral Images.

Fen Chen,Ting Feng Tang,Ke Wang

DOI: https://doi.org/10.1109/LGRS.2015.2504426

IF: 5.343

2016-01-01

IEEE Geoscience and Remote Sensing Letters

Abstract:In many hyperspectral image processing tasks such as classification, unmixing, and target detection, jointly using the pixels in an image patch can generally improve the performance. In this letter, we propose using a low-rank decomposition model to analyze the image patch. The image patch is decomposed into the sum of a low-rank matrix, a sparse matrix, and a bounded matrix using a convex optimiz...