Fig. 1. The process of making the optical lens. (a) Self-designed single glued lens; (b) PSF of lens at 20° FOV

Fig. 2. Deep iterative adaptive convolution

Fig. 3. Deblurring network with DIFAN

Fig. 4. Comparison of different restoration methods based on simulation dataset. (a) Original blurred image; (b) recovery results of ISD-Deblur; (c) recovery results of DeblurGAN-v2 ; (d) recovery results of proposed DIFAN; (e) ground truth images

Fig. 5. Comparison of network restoration effects based on simulation datasets at different field angles

Fig. 6. Network restoration effect of simulation data set with a plano-convex lens at 15° field of view angle

Fig. 7. Distortion correction and display-capture experiments

Fig. 8. Image pairs taken using display capture devices

Fig. 9. Experimental results. (a) The original blurred images; (b) image recovery results from the network; (c) ground truth images