Fig. 1. Optical configuration of LI-COACH

Fig. 2. G-S algorithm generates bCPM

Fig. 3. Plots of PSNR and the DC bias value versus the number of iterations. (a) DBSA sequentially selects bCPM iterations; (b) DBSRT alternately selects bCPM iterations

Fig. 4. Reconstructed images of bCPMs. (a) Reconstructed image by initial bCPMs; (b) reconstructed image after DBSA; (c) reconstructed image after DBSRT for alternating strategies

Fig. 5. PBR varying with the number of iterations

Fig. 6. Experimental setup of LI-COACH

Fig. 7. Experimental data. (a1)(b1) Binary CPMs: bCPM1 and bCPM2; (a2)(b2) PSF of initial bCPM; (a3)(b3) object intensity response of initial bCPM; (a4)(b4) PSF of optimized bCPM; (a5)(b5) object intensity response of optimized bCPM

Fig. 8. Experimental reconstruction images. (a) PSH; (b) hologram of the object; (c) initial bCPMs reconstruction image; (d) regular image; (e) reconstructed image after DBSA optimization; (f) reconstructed image after DBSRT for alternating strategies; (g) reconstruction image of bCPM1 optimized by DBSA and initial bCPM2