Fig. 1. Basic optical path of CMI algorithm, phase distribution of adopted binary random phase plate, and representative diffraction pattern. (a) Basic optical path of CMI algorithm; (b) phase distribution of adopted binary random phase plate; (c) representative diffraction pattern

Fig. 2. Relationship between B and φ(x) varying with increasing iterations

Fig. 3. Matrix representation of CMI pattern generation

Fig. 4. Scheme of convolution

Fig. 5. Basic optical path and simulation results. (a) Amplitude and phase of sample; (b) incident light of modulation plate; (c) spectrum and phase map of modulation plate; (d) pattern recorded by detector without modulation plate (image of sample); (e) diffraction pattern with modulation plate; (f) basic optical path

Fig. 6. Reconstructed results with different number of equations. (a) Complete diffraction pattern; (b)--(e) pattern random sampling when N_ratio is 0.2, 1, 2, and 5, respectively; (f)--(j) reconstructed results corresponding to Figs. 6(a)--(e)

Fig. 7. Error curve and corresponding error gradient curve under different number of equations

Fig. 8. Reconstructed results under different spectral width of modulation plate. (a)--(e) Spectra of different modulators and corresponding phase maps; (f)--(j) diffraction patterns corresponding to Figs. 8(a)--(e); (k)--(o) reconstructed results corresponding to Figs. 8(a)--(e)

Fig. 9. Error curves and corresponding error gradient curves of reconstructed results with different spectral width of modulation plate

Fig. 10. Reconstruction results of light to be measured with different sparsity characteristics and same spectrum of modulation plate. (a)--(d) Diffraction patterns under different N″; (e)--(h) reconstructed results corresponding to Figs. 10(a)--(d)

Fig. 11. Reconstructed error under different spot width of light to be measured, without modulation plate. Green dashed line represents critical width