Joint Collaborative Representation With Shape Adaptive Region and Locally Adaptive Dictionary for Hyperspectral Image Classification

• Published in: IEEE geoscience and remote sensing letters Vol. 17; no. 4; pp. 671 - 675
• Main Authors: Yang, Jinghui, Qian, Jinxi
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.04.2020; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Adaptive dictionary; Adaptive filters; Classification; Collaboration; Dictionaries; Glossaries; hyperspectral image (HSI) classification; Hyperspectral imaging; Image classification; Image filters; joint collaborative representation; Pixels; Representations; Shape; shape adaptive (SA); Spatial data; Support vector machines; Training
• ISSN: 1545-598X; 1558-0571
• DOI: 10.1109/LGRS.2019.2929840

A novel hyperspectral image (HSI) classification method based on joint collaborative representation with shape adaptive region and locally adaptive dictionary (SALJCR) is proposed in this letter. First, the shape adaptive (SA) region is selected for each pixel to exploit the neighboring spatial information adaptively. The average filtering (according to SA regions) is performed for the whole image. Then, based on the filtered image, a locally adaptive dictionary is constructed for each test pixel to reduce the negative impact of irrelevant pixels on representation. Finally, a joint collaborative representation method is applied to decompose the pixels and assign the class label. Experimental results demonstrate that the proposed SALJCR method outperforms some state-of-the-art classifiers.