CRCOED: Collaborative Representation-Based Classification using Odd Even Decomposition for Hyperspectral Remote Sensing Imagery

In the modern era exploiting the texture data is of excessive concern for classification of hyperspectral imagery (HSI). However, it is very challenging and prolonged task to obtain appropriate training samples which represents most discriminative features. This research paper suggested an improved collaborative representation based classification by means of odd even decomposition theorem i.e. CRCOED. The proposed method employs that augment the training samples using odd even decomposition theorem. In addition, every sample is characterized by means of a linear combination of the training samples from the whole training set; after that reconstruction of the image will be performed with related involvement from each class. The main purpose of the suggested method is to produce the highest parity symmetrical demonstration of sample for accurate and robust classification. Lastly, the observed outcomes on numerous HSI data sets revealed that this technique is more effective in comparison to recent methods. (C) 2018 The Authors. Published by Elsevier B.V.