The reconstruction of high resolution image based on compressed sensing

Constrained by traditional sampling theory, it is difficult to obtain high resolution image directly by signal acquisition system. The method that uses compressed sensing technology to measure the high resolution image and reconstruct with measurements breaks the bottleneck of Nyquist sampling theory, which is a new application for compressed sensing in image processing field. In CS, the reconstruction of super resolution image can be converted to how to construct measurement matrix and design reconstruction algorithm. For the reason that Gaussian measurement matrix requires a great number of high dimensional projection computations, we introduce sparse random projection into compressed sensing, proposing a measurement matrix which obeys sparse random projection distribution: sparse projection matrix. For the reason that the existing OMP algorithms require a lot of linear measurements to ensure accurate reconstruction, we propose an improved OMP algorithm. Experimental results show that, with sparse projection matrix and the improved OMP algorithm, high resolution image can be reconstructed accurately with little number of measurements.