ABSTRACT
Compressive sensing is a well-known process for the reconstruction of Hyperspectral images (HSIs). Compressive sensing based techniques make use of spatial or spectral information in the HSIs. Direct reconstruction of HSI causes higher order complexity at the computational process and also causes difficulty in estimation of abundance and end members. In order to overcome the above mentioned drawbacks during reconstruction phase, in this paper a compressive sensing based technique called data adaptable sparse reconstruction (DASR) technique has been proposed. Here, data adaptable parameter and TV (Total Variation) Regularizer is combined under the Unmixing process to get the necessary HSI data characteristics. TV regularizer is considered to offer smooth abundance under several discontinuities. But it is not adequate to estimate the end-member function. Therefore, the data adaptable parameter is proposed that minimizes the mixture of pixels by maximizing the abundance maps sparsity. To validate the performance of the proposed model, it is compared with the existing techniques.
