Search

EP-4738868-A1 - DIFFRACTION-GRATING-BASED DEPTH IMAGING SYSTEMS AND METHODS WITH PIXEL CROSSTALK MITIGATION

EP4738868A1EP 4738868 A1EP4738868 A1EP 4738868A1EP-4738868-A1

Abstract

Diffraction-grating-based depth imaging systems and methods for pixel crosstalk mitigation are disclosed. The system includes a transmissive diffraction mask (TDM) and an underlying image sensor. The TDM receives light from a scene and generates diffracted light encoding angle-of-incidence information, which is detected by the image sensor comprising a plurality of pixels. The TDM has a grating structure that spatially maps the diffracted light onto distinct pixel groups, including bright-positive pixels that receive a portion of the diffracted light and exhibit a pixel response that increases with angle of incidence, bright-negative pixels that receive another portion of the diffracted light and exhibit a pixel response that decreases with angle of incidence, and dark pixels that receive substantially no diffracted light. The dark pixels are interspersed among the bright-positive and bright-negative pixels to mitigate pixel crosstalk, enhancing the accuracy and contrast of depth measurements.

Inventors

  • FAVRON, Alexandre
  • FAGHIHI, Niloufar
  • GREGOIRE, PASCAL
  • SUMMY, Simon

Assignees

  • Airy3d Inc.

Dates

Publication Date
20260506
Application Date
20251028

Claims (15)

  1. A depth imaging system (100), comprising: a transmissive diffraction mask (TDM) (108) configured to receive light (102) from a scene (104) and generate diffracted light (110) encoding angle-of-incidence information; and an image sensor (112) positioned to detect the diffracted light (110) generated by the TDM (108) and comprising a plurality of pixels (128), wherein the TDM (108) has a grating structure configured to spatially map the diffracted light (110) onto distinct pixel groups of the plurality of pixels (128), the distinct pixel groups comprising: (i) bright-positive pixels (+) that receive a portion of the diffracted light (110) from the TDM (108) and exhibit a pixel response that increases with increasing angle of incidence; (ii) bright-negative pixels (-) that receive another portion of the diffracted light (110) from the TDM (108) and exhibit a pixel response that decreases with increasing angle of incidence; and (iii) dark pixels (D) that receive substantially no diffracted light (110) from the TDM (108) and exhibit substantially no pixel response, the dark pixels (D) being interspersed among the bright-positive pixels (+) and the bright-negative pixels (-) to reduce pixel crosstalk within the image sensor (112).
  2. The depth imaging system (100) of claim 1, further comprising a computer device (114) operatively coupled to the image sensor (112) and configured to determine depth information about the scene (104) based on differences between the pixel responses of the bright-positive (+) and bright-negative (-) pixels.
  3. The depth imaging system (100) of claim 1 or 2, wherein the grating structure of the TDM (108) is configured to partition the plurality of pixels (128) according to a design rule specifying an overall pixel group distribution of 25% bright-positive pixels (+), 25% bright-negative pixels (-), and 50% dark pixels (D), and wherein the design rule for the overall pixel group distribution is enforced locally at a scale of four-pixel clusters.
  4. The depth imaging system (100) of claim 1 or 2, wherein the grating structure is configured to partition the plurality of pixels (128) according to a design rule specifying that within any given 2×2 pixel cluster, there are two dark pixels (D) and two bright pixels of either the same or different polarities.
  5. The depth imaging system (100) of claim 4, wherein: the TDM (108) comprises a grating axis (108) extending within a plane of incidence (142) in which the bright-positive (+) and bright-negative (-) pixels are sensitive to angle of incidence; and the grating structure is configured to partition the plurality of pixels (128) according to an additional design rule specifying: (a) if the two dark pixels (D) in the given 2×2 pixel cluster are aligned along a direction perpendicular to the grating axis (118), the two bright pixels within the given 2×2 pixel cluster are either both bright-positive (+) or both bright-negative (-); and (b) if the two dark pixels (D) in the given 2×2 pixel cluster are aligned along a direction not perpendicular to the grating axis (118), the two bright pixels within the given 2×2 pixel cluster consist of one bright-positive pixel (+) and one bright-negative pixel (-).
  6. The depth imaging system (100) of any one of claims 1 to 5, wherein the TDM (108) comprises one or more mask layers (144) disposed over the image sensor (112).
  7. The depth imaging system (100) of claim 6, wherein at least one of the one or more mask layers (144) comprises an array of blocks (146) arranged on a base substrate (148), and wherein each block (146) is centered over a boundary between a bright-positive pixel (+) and a bright-negative pixel (-).
  8. The depth imaging system (100) of claim 7, wherein each block (146) comprises a lower prism (150) and an upper prism (152), with the lower prism (150) having a larger base to form a stepped structure, and wherein the lower (150) and upper (152) prisms of each block (146) are preferably hexagonal or rectangular.
  9. A method (200) for depth imaging, comprising: diffracting (202) light (102) received from a scene (104) with a transmissive diffraction mask (TDM) to generate diffracted light (110) encoding angle-of-incidence information; detecting (204) the diffracted light (110) with an image sensor (112) comprising a plurality of pixels (128), wherein the TDM (108) spatially maps the diffracted light (110) onto distinct pixel groups of the plurality of pixels (128), the distinct pixel groups comprising: (i) bright-positive pixels (+) that receive a portion of the diffracted light (110) and exhibit a pixel response that increases with increasing angle of incidence; (ii) bright-negative pixels (-) that receive another portion of the diffracted light (110) and exhibit a pixel response that decreases with increasing angle of incidence; and (iii) dark pixels (D) that receive substantially no diffracted light (110) and exhibit substantially no pixel response, the dark pixels (D) being interspersed among the bright-positive pixels (+) and the bright-negative pixels (-) to reduce pixel crosstalk within the image sensor (112); and determining (206) depth information about the scene (104) based on differences between the pixel responses of the bright-positive (+) and bright-negative (-) pixels.
  10. The method (200) of claim 9, wherein the TDM (108) partitions the plurality of pixels (128) according to a design rule specifying an overall pixel group distribution of 25% bright-positive pixels (+), 25% bright-negative pixels (-), and 50% dark pixels (D), and wherein the design rule for the overall pixel group distribution is enforced locally at a scale of four-pixel clusters.
  11. The method (200) of claim 9, wherein the TDM (108) partitions the plurality of pixels (128) according to a design rule specifying that within any given 2×2 pixel cluster, there are two dark pixels (D) and two bright pixels of either the same or different polarities.
  12. The method (200) of claim 11, wherein: the TDM (108) comprises a grating axis (118) extending within a plane of incidence (124) in which the bright-positive (+) and bright-negative (-) pixels are sensitive to angle of incidence; and the TDM (108) partitions the plurality of pixels (128) according to an additional design rule specifying: (a) if the two dark pixels in the given 2×2 pixel cluster are aligned along a direction perpendicular to the grating axis (118), the two bright pixels within the given 2×2 pixel cluster are either both bright-positive (+) or both bright-negative (-); and (b) if the two dark pixels in the given 2×2 pixel cluster are aligned along a direction not perpendicular to the grating axis (118), the two bright pixels within the given 2×2 pixel cluster consist of one bright-positive pixel (+) and one bright-negative pixel (-).
  13. A transmissive diffraction mask (TDM) (108) for mitigating pixel crosstalk in depth imaging, comprising a grating structure configured to diffract light (102) received from a scene (104) by encoding angle-of-incidence information and directing the diffracted light (110) onto an image sensor (112) comprising a plurality of pixels (128), wherein the grating structure is further configured to spatially map the diffracted light (112) onto distinct pixel groups of the plurality of pixels (128), the distinct pixel groups comprising: (i) bright-positive pixels (+) onto which the TDM (108) is configured to direct a portion of the diffracted light (110), the bright-positive pixels (+) exhibiting a pixel response that increases with increasing angle of incidence; (ii) bright-negative pixels (-) onto which the TDM (108) is configured to direct another portion of the diffracted light (110), the bright-negative pixels (-) exhibiting a pixel response that decreases with increasing angle of incidence; and (iii) dark pixels (D) onto which the TDM (108) is configured to direct substantially no diffracted light (110), the dark pixels (D) exhibiting substantially no pixel response and being interspersed among the bright-positive pixels (+) and the bright-negative pixels (-) to reduce pixel crosstalk within the image sensor (112).
  14. The TDM (108) of claim 13, wherein the grating structure is configured to partition the plurality of pixels (128) according to a design rule specifying an overall pixel group distribution of 25% bright-positive pixels (+), 25% bright-negative pixels (-), and 50% dark pixels (D), the design rule being enforced locally at a scale of four-pixel clusters.
  15. The TDM (108) of claim 13, wherein the TDM (108) comprises a grating axis (118) extending within a plane of incidence (142) in which the bright-positive (+) and bright-negative (-) pixels are sensitive to angle of incidence, and wherein the grating structure is configured to partition the plurality of pixels (128) according to: a first design rule specifying that within any given 2×2 pixel cluster, there are two dark pixels (D) and two bright pixels of either the same or different polarities; and a second design rule specifying: (a) if the two dark pixels (D) in the given 2×2 pixel cluster are aligned along a direction perpendicular to the grating axis (118), the two bright pixels within the given 2×2 pixel cluster are either both bright-positive (+) or both bright-negative (-); and (b) if the two dark pixels (D) in the given 2×2 pixel cluster are aligned along a direction not perpendicular to the grating axis (118), the two bright pixels within the given 2×2 pixel cluster consist of one bright-positive pixel (+) and one bright-negative pixel (-).

Description

TECHNICAL FIELD The present disclosure relates to imaging technology, specifically focusing on techniques for reducing pixel-to-pixel crosstalk in depth imaging systems. BACKGROUND Pixel-to-pixel crosstalk is a well-documented issue in image sensors that utilize arrays of photosensitive pixels, including complementary metal-oxide-semiconductor (CMOS) and charge-coupled device (CCD) architectures. Crosstalk occurs when light intended for a specific pixel is not fully absorbed by the pixel and leaks into neighboring pixels, resulting in unwanted signals and a degradation of image quality. Several factors can influence the extent of crosstalk, including pixel size, design, material composition, and the wavelength of incoming light. Crosstalk tends to increase with longer wavelengths, such as infrared light, due to their greater penetration depth in semiconductor materials like silicon. Mitigating pixel-to-pixel crosstalk in various imaging applications remains an ongoing challenge. SUMMARY The present disclosure pertains to techniques for reducing pixel-to-pixel crosstalk in diffraction-grating-based depth imaging systems. These systems incorporate a diffraction grating, referred to herein as a "transmissive diffraction mask" (TDM), placed in front of an image sensor to create pixels with angle-sensitive responses. Utilizing near-field diffraction effects, specifically the Talbot effect, the TDM encodes directional information into the light intensity distribution received from a scene, which is then transmitted as diffracted light to the image sensor beneath it. Consequently, the pixels can measure both the intensity and the angle of incoming light, allowing for the determination of depth information about the scene. However, crosstalk in these systems can distort the angular response of the pixels, thereby reducing the accuracy of depth measurements, particularly with longer wavelengths such as infrared. In accordance with an aspect, there is provided a depth imaging system, comprising: a transmissive diffraction mask (TDM) configured to receive light from a scene and generate diffracted light encoding angle-of-incidence information; andan image sensor positioned to detect the diffracted light generated by the TDM and comprising a plurality of pixels,wherein the TDM has a grating structure configured to spatially map the diffracted light onto distinct pixel groups of the plurality of pixels, the distinct pixel groups comprising: (i) bright-positive pixels that receive a portion of the diffracted light from the TDM and exhibit a pixel response that increases with increasing angle of incidence; (ii) bright-negative pixels that receive another portion of the diffracted light from the TDM and exhibit a pixel response that decreases with increasing angle of incidence; and (iii) dark pixels that receive substantially no diffracted light from the TDM and exhibit substantially no pixel response, the dark pixels being interspersed among the bright-positive pixels and the bright-negative pixels to reduce pixel crosstalk within the image sensor. In some embodiments of the depth imaging system, the depth imaging system further comprises a computer device operatively coupled to the image sensor and configured to determine depth information about the scene based on differences between the pixel responses of the bright-positive and bright-negative pixels. In some embodiments of the depth imaging system, the grating structure of the TDM is configured to partition the plurality of pixels according to a design rule specifying an overall pixel group distribution of 25% bright-positive pixels, 25% bright-negative pixels, and 50% dark pixels. In some instances, the design rule for the overall pixel group distribution is enforced locally at a scale of four-pixel clusters. In some embodiments of the depth imaging system, the grating structure is configured to partition the plurality of pixels according to a design rule specifying that within any given 2×2 pixel cluster, there are two dark pixels and two bright pixels of either the same or different polarities. In some instances, the TDM comprises a grating axis extending within a plane of incidence in which the bright-positive and bright-negative pixels are sensitive to angle of incidence; and the grating structure is configured to partition the plurality of pixels according to an additional design rule specifying: (a) if the two dark pixels in the given 2×2 pixel cluster are aligned along a direction perpendicular to the grating axis, the two bright pixels within the given 2×2 pixel cluster are either both bright-positive or both bright-negative; and(b) if the two dark pixels in the given 2×2 pixel cluster are aligned along a direction not perpendicular to the grating axis, the two bright pixels within the given 2×2 pixel cluster consist of one bright-positive pixel and one bright-negative pixel. In some embodiments of the depth imaging system, the TDM comprises one or more mask