The reconstruction results obtained by FLBM has slightly better appearances than ALBM, FLBI, LBM and BP, and has better appearances than OMP.

Table 2 gives the quantitative results of the FLBM, ALBM, FLBI, LBM, BP and OMP algorithms.

However, the FLBM is significantly superior to the LBM, BP and OMP algorithm, and is slightly superior to the ALBM and FLBI algorithm, for the same sampling rate.

To confirm the universality of the proposed FLBM algorithm, we apply it now to reconstruct the three different groups of the test images.

To illustrate the FLBM robust to noise, a zero-mean

The proposed FLBM reconstruction algorithm is applied to these test images with Gaussian

5 (b), (c), (d), (e), (f) and (g) are the reconstructed Lena images obtained by the FLBM, ALBM, FLBI, LBM, BP and OMP, respectively.

To confirm the robustness of the proposed FLBM algorithm, the PSNR (dB) is used to measure the performance of the proposed algorithm for the noisy Lena image (256 x 256).

Table 5 gives the PSNR of the reconstructed Lena image at different noise levels resulting from the FLBM, ALBM, FLBI, LBM, BP and OMP algorithms.

For further comparison, the PSNR in dB of the reconstructed different noisy images at different noise levels resulting from the FLBM and ALBM algorithms are listed in Table 6.

An effective and fast algorithm, referred to as FLBM, has been proposed to reconstruct the sparse coefficients from the random measurement, thereby to reconstruct the signal.