eess.IVApr 11, 2026

Compact single-shot ranging and near-far imaging using metasurfaces

Authors: Junjie Luo, Yuxuan Liu, Wei Ting Chen, Qing Wang, Qi Guo

Organizations: Purdue University, 501 Northwestern Ave, West Lafayette, IN, 47906 · SNOChip INC., 98 Marion Dr, Plainsboro, NJ, 08536

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

We present a metasurface imaging system capable of simultaneously capturing two images at close range (1-2cm) and an additional image at long range (about 40cm) on a shared photosensor. The close-range image pair focuses at 1.4cm and 2.0cm, respectively, which forms a focal stack, enabling passive ranging with an accuracy of ±\pm1mm from 12mm to 20mm through a computationally efficient depth-from-defocus algorithm for a simplified scenario. The entire system is compact, with a total track length of 15mm, making it suitable for seamless integration into edge platforms for defense and other resource-constrained applications.

Figures & tables

Figure 1: Schematic of the proposed imaging system. The front layer consists of three laterally spaced apertures, where the two side panels incorporate metasurface phase profiles that introduce both angular deflection and additional optical powers ρ1\rho_{1} and ρ3\rho_{3} , while the central panel remains unmodulated. The three beams are subsequently focused by a shared refractive lens with an optical power ρL\rho_{L} onto a common photosensor, forming three spatially multiplexed sub-images I1−3I_{1-3} . Due to the different effective optical powers ( ρ1+ρL\rho_{1}+\rho_{L} , ρL\rho_{L} , and ρ3+ρL\rho_{3}+\rho_{L} ), the three sub-images correspond to distinct focal plane distances Zf,1Z_{f,1} – Zf,3Z_{f,3} . An aperture stop is used to spatially separate the image regions and prevent overlap. Distances s1s_{1} and s2s_{2} denote the separations between optical elements and the sensor, respectively.
Figure 2: Prototype of the optical system illustrated in Fig. 1 . (a) The first-layer optical element and the SEM images of the designed nano-structure. (b) The entire system after assembly. The total track length of the system is only 15 mm.
Figure 3: Sample measurement captured by the prototype. A microscope slide and a plastic ruler are placed approximately 1 cm and 40 cm in front of the system, respectively. Consistent with the optical layout in Fig. 1 , the system forms three side-by-side images, I1−3I_{1-3} , on a shared image sensor. The images I1I_{1} and I3I_{3} are focused on the close-range target with slightly different focal planes and provide sub-millimeter spatial resolution, while I2I_{2} is focused on the far-range target. As a result, the close-range target is completely blurred out in I2I_{2} , allowing the far-range target to be observed clearly.
Figure 4: Measured point spread functions of close-range sub-images I1I_{1} and I3I_{3} . We capture the image of a point source placed from 1212 mm to 2020 mm with an interval of 11 mm
Figure 5: Depth estimation using the prototype. (a) The frequencies of estimated depth values for each true depth using Eq. 1 and image pairs in Fig. 4 as input. The dotted lines indicate of 5%5\% relative error tolerance. (b) Mean and deviation of the estimated depth values. The solid red curves indicate the mean predicted depth, and the half-widths of the color bands represent the mean absolute error (MAE) in prediction at each true depth. The black solid lines are boundaries of 5%5\% relative error. The MAE is constantly smaller than 1 mm within the working range of 12 mm to 20 mm.
More in this topic

Depth Estimation

Depth estimation infers scene distance from images and sensors, enabling 3D reconstruction, navigation, and spatial understanding.

See all 198 papers

Explore similar work

CardsList

  1. Depth from Dual Differential Defocus and Stereo Consensus

    Jun 1, 2026Junjie Luo, Wei Xu, Dylan Chu +2Metric Depth EstimationDepth Estimation

    Purdue University, West Lafayette, IN, USA · Northwestern University, Evanston, IL, USA

  2. Broadband Mid-Wave Infrared Imaging Enabled by All-Silicon Hyperboloid Metalens

    Oct 21, 2026Rose Johnson, Debabrata Mandal, Zihao Zou +6Computational ImagingImage Reconstruction

    Department of Electrical and Computer Engineering, University of Washington, Seattle, WA 98195, USA. · Department of Computer Science, University of North Carolina at Chapel Hill, Chapel Hill, NC 27517, USA. · Department of Physics, University of Washington, Seattle, WA 98195, USA.

  3. Comparative Study of Vision-Based Metric Measurement for Large-Scale Planar Scenes

    May 26, 2026ZhiXin SunStereo Vision

    PowerChina Zhongnan Engineering Corporation Limited