Fig. 1. Framework of using a monocular metasurface camera for single-shot 4D imaging.

a The target scene consists of objects with different polarization states and depths. b Light emitted from the scene is split and focused on the photosensor, forming two images of orthogonal linear polarizations. The red and blue colors correspond to x- and y-polarized light, respectively. c The raw encoded image-pair recorded on the photosensor, with depth information of the scene encoded in the local orientation of the translation vector (green arrows). The black outlines are the geometric image positions of the objects. d Decoded 4D image of the target scene, including all-in-focus 2D light intensity (I_x and I_y), polarization contrast (I_x/I_y), and depth (z_obj), can be retrieved from the raw image-pair by a straightforward, physically interpretable algorithm based on image segmentation and template matching of the object-pair.