Performance evaluation of compressive sensing matching pursuit backtracking iterative hard thresholding algorithm for improving reconstruction

Now-a-days due to advancements in technologies most of the applications in signal processing were using the models based on the sparse signal. Sub optimal strategies were used in these models to estimate the sparsest coefficients. In this work various algorithms were analyzed to address its optimal solutions. The sparsest solution can be found for the linear equations which are under determined. In this work, a complete study is carried out based on Compressive Sensing Matching Pursuit Back Tracking Iterative Hard Threshold (CMPBIHT) algorithm in the real-world scenario. As the BIHT algorithm may often fail to converge and its performance seems to be degraded if the conditions fail. To address these challenges, we have modified the BIHT algorithm to guarantee the convergence using the proposed method, even in this regime. Further the proposed CMPBIHT algorithm is evaluated and compared with the state of art techniques and it is observed that the proposed algorithm retains the similarities of the original algorithm. In this proposed model we have adopted the Compressive Sensing (CS) schemes along with Orthogonal Matching Pursuit (OMP). With this proposal we are able to solve the least squares problem for the new residual. We also investigated the reliability in sparse solutions along with compressive sensing techniques while decoding and over complete representations. An extensive research is carried out at the reconstruction side with the fundamental theme of CS, IHT and OMP techniques. The simulation results perform better efficiency at the reconstruction of the Gaussians signals by guaranteeing the productions in the residual error and noise. Further the proposed algorithm performs better at the reconstruction with nominal complexity in each of the iteration computationally.