Abstract: Hyperspectral imagery provide accurate and detailed information extraction than possible with any other type of remotely sensed data. This improvement comes with computational complexity and over dimensionality. There is an increase in interest in dimensionality reduction through random projections due in part to the emerging paradigm of compressed sensing. CS exploits the fact that many signals are sparse in the sense that they have concise representation in certain basis called dictionary. Traditionally some transform based fixed dictionaries such as DFT, DCT, DWT are used which are relatively easy to analyse. But they are over simplistic and for certain real time data such as hyperspectral images they offer less accuracy. An approach that has been recently proven to be very effective is adaptive dictionary learning technique in which the dictionary is constructed adaptively using the input image for better sparsity. While learning the dictionary the most important computational challenge is the solution of corresponding optimization problem. The reconstruction strategies like compressive projection principal component analysis, multi-hypothesis prediction method and several class dependent strategies were proposed for the reconstruction of hyperspectral imagery from random projections. In this paper, a brief review of various dictionary learning methods and reconstruction techniques for HSI is presented along with their performance evaluation.

Keywords: Hyper Spectral Images, Compressive Sensing, Principal Component Analysis (PCA), Multi Hypothesis Prediction.