Point spread function for diffuser cameras based on wave propagation and projection model.

Compared to traditional imaging system, diffuser camera is an easy-built imaging system to capture the light field with small form factor. Its imaging target can be reconstructed by deconvolving the sensor images with the point-spread-function (PSF) at the corresponding depth. However, the existing method to obtain the PSFs is generally relied on measuring point-source-responses at several depths which presents high complexity and low reliability. In this paper, we propose a theoretical PSF model for the diffuser camera by estimating the diffuser's phase based on projection model and deriving the image response at any depth by forward Fourier optics to enable the reconstruction of the object at any depth. The experimental results demonstrate the effectiveness of the proposed model in terms of the correlation between the captured PSFs and our model derived PSFs, the correlation between the ground truth image and reconstructed images under different depths, and the correlation between the images reconstructed by real captured PSFs and the images reconstructed by our model derived PSFs. The objects at different depths can be correctly reconstructed by our theoretically derived PSFs, which benefits the application of diffuser camera for much lower complexity.