IP Library › Granted Patent US 12,423,974
Granted Patent B2
US 12,423,974 · App. 18/375,071 · Granted Sep 23, 2025

Vision sensor, event signal processing method thereof, and image processing device including the same

Inventors: Keunjoo Park (Suwon-si, KR); Junseok Kim (Suwon-si, KR); Junhyuk Park (Suwon-si, KR); Bongki Son (Suwon-si, KR)
Assignee: SAMSUNG ELECTRONICS CO., LTD.
G06V20/44G06V10/60
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,423,974
App. No.
18/375,071
Granted
Sep 23, 2025
Kind
B2
Abstract

A vision sensor includes a memory; a pixel array including a plurality of pixels configured to generate an event signal by detecting an event in which an intensity of incident light changes; and an event detection circuit configured to read the event signal from the pixel array and configured to process the event signal, wherein the event detection circuit configured to: store, in the memory, a state map including a first state corresponding to a presence of an event of each of the plurality of pixels, and a second state corresponding to an absence of the event, and correct the event occurring in a selected pixel, based on the state map stored in the memory.

Claims (67)

1. A vision sensor comprising:

a memory;

a pixel array comprising a plurality of pixels configured to generate an event signal by detecting an event in which an intensity of incident light changes; and

an event detection circuit configured to read the event signal from the pixel array and configured to process the event signal,

wherein the event detection circuit is configured to:

store, in the memory, a state map comprising:

a first state corresponding to a presence of an event of each of the plurality of pixels, and

a second state corresponding to an absence of the event, and

correct the event occurring in a selected pixel, based on the state map stored in the memory,

wherein the pixel array comprises a target pixel and surrounding pixels adjacent to the target pixel,

wherein the event detection circuit is configured to determine the target pixel to be a hot pixel, based on the number of the first state acquired from the surrounding pixels, which is less than a preset hot pixel reference value,

wherein the event detection circuit is configured to determine the target pixel to be a normal pixel, based on the number of the first state acquired from the surrounding pixels, which is greater than or equal to the preset hot pixel reference value,

wherein the event detection circuit is configured to reset the state map stored in the memory at a predetermined time interval, and

wherein the preset hot pixel reference value is inversely proportional to the predetermined time interval.

2. The vision sensor of claim 1 , wherein the event detection circuit is configured to determine the target pixel to be the hot pixel based on the event signal corresponding to a second state.

3. The vision sensor of claim 1 ,

wherein the event detection circuit is configured to determine the target pixel to be a dead pixel, based on the number of the first state acquired from the surrounding pixels arranged in one direction penetrating the target pixel, which is greater than a preset dead pixel reference value, and based on the target pixel that is in the second state.

4. The vision sensor of claim 3 , wherein the event detection circuit is configured to acquire the number of the first state acquired from the surrounding pixels arranged in one direction penetrating the target pixel in each direction from a first direction to an n-th direction, and

wherein the event detection circuit is configured to compare a maximum number among numbers of the first state acquired from the surrounding pixels arranged in one direction penetrating the target pixel in the first direction to the n-th direction with the preset dead pixel reference value.

5. The vision sensor of claim 3 , wherein the event detection circuit is configured to reset the state map stored in the memory at a predetermined time interval, and

wherein the preset dead pixel reference value is proportional to the predetermined time interval.

6. The vision sensor of claim 3 , wherein the event detection circuit is configured to determine the target pixel to be the dead pixel by the event signal corresponding to the first state.

7. The vision sensor of claim 1 , wherein the memory comprises a same number of flip-flops as a number of the plurality of pixels.

8. The vision sensor of claim 1 , wherein the event detection circuit is configured to scan the plurality of pixels in units of columns to read the event signal, and

wherein the event detection circuit is configured to store an event state for each of the plurality of pixels arranged in three consecutive columns in the memory.

9. The vision sensor of claim 8 , wherein the plurality of pixels are arranged in a matrix of i rows and j columns, and

wherein the memory comprises 3i flip-flops.

10. A method performed by a vision sensor, the method comprising:

detecting an event in which an intensity of incident light changes and reading an event signal from a pixel array comprising a plurality of pixels configured to generate an event signal;

storing, in a memory, a state map comprising:

a first state corresponding to a presence of an event of each of the plurality of pixels, and

a second state corresponding to an absence of the event; and

correcting the event occurring in a selected pixel based on the state map stored in the memory,

wherein the pixel array comprises a target pixel and surrounding pixels adjacent to the target pixel,

wherein the correcting the event occurring in the selected pixel based on the state map stored in the memory, comprises: determining hot pixel

wherein the determining hot pixel comprises:

acquiring a number of the first state from the surrounding pixels, and

comparing the number of the first state with a preset hot pixel reference value resetting the state map stored in the memory at predetermined time interval,

wherein, based on the number of the first state, which is less than the preset hot pixel reference value, the target pixel is determined to be a hot pixel, and

wherein, based on the number of the first state, which is greater than or equal to the preset hot pixel reference value, the target pixel is determined to be a normal pixel

wherein the preset hot pixel reference value is inversely proportional to the predetermined time interval.

11. The method of claim 10 , further comprising determining the target pixel to be the hot pixel by an event signal corresponding to a second state.

12. The method of claim 10 ,

wherein the correcting the event occurring in the selected pixel based on the state map stored in the memory, further comprises determining dead pixel

wherein the determining dead pixel comprises:

acquiring a number of the first state for the surrounding pixels arranged in one direction penetrating the target pixel, and

comparing the number of the first state with a preset dead pixel reference value, and

wherein the target pixel is determined to be a dead pixel, based on the number of the first state, which is greater than the preset dead pixel reference value and based on the target pixel that is in the second state.

13. The method of claim 12 , wherein the determining dead pixel comprises resetting the state map stored in the memory at a predetermined time interval,

wherein the preset dead pixel reference value is proportional to the predetermined time interval.

14. The method of claim 12 , wherein the determining dead pixel comprises:

acquiring the number of the first state in each direction from a first direction to an n-th direction, and

comparing a maximum number among numbers of the first state in the first direction to the n-th direction with the preset dead pixel reference value.

15. The method of claim 13 , further comprising determining the target pixel to be a dead pixel by an event signal corresponding to the first state.

16. An image processing device comprising a vision sensor and an image sensor, wherein the vision sensor comprises:

a pixel array comprising a plurality of pixels configured to generate an event signal by detecting an event in which an intensity of incident light changes; and

an event detection circuit configured to read the event signal from the pixel array and configured to process the event signal,

wherein the event detection circuit is configured to store, in a memory, a state map comprising:

a first state corresponding to a presence of an event of each of the plurality of pixels, and

a second state corresponding to an absence of an event, and

wherein the event detection circuit is configured to correct an event occurring in a selected pixel, based on the state map stored in the memory,

wherein the pixel array comprises a target pixel and surrounding pixels adjacent to the target pixel,

wherein the event detection circuit is configured to acquire a number of the first state from the surrounding pixels,

wherein the event detection circuit is configured to determine the target pixel to be a hot pixel, based on the number of the first state, which is less than a preset hot pixel reference value,

wherein the event detection circuit is configured to determine the target pixel to be a normal pixel, based on the number of the first state, which is greater than or equal to the preset reference hot pixel value,

wherein the event detection circuit is configured to reset the state map stored in the memory at a predetermined time interval, and

wherein the preset hot pixel reference value is inversely proportional to the predetermined time interval.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Sep 29, 2023
From: PARK, KEUNJOO; KIM, JUNSEOK; PARK, JUNHYUK; SON, BONGKI
To: SAMSUNG ELECTRONICS CO., LTD.
Reel/Frame 065078/0202 →
Priority Claims (1)
KR 10-2023-0017878 · Feb 10, 2023 · national
Continuity (1)
Related Publication 20240273901A1 · Aug 15, 2024
References Cited (17)
US 9781365B2 · Subbotin · 2017 [cited by examiner]
US 9817507B2 · Park et al. · 2017 [cited by applicant]
US 11202025B2 · Park et al. · 2021 [cited by applicant]
US 11405567B2 · Seo · 2022 [cited by applicant]
US 20030063202A1 · Toyoda · 2003 [cited by examiner]
US 20070040920A1 · Kinoshita · 2007 [cited by examiner]
US 20070291142A1 · Noh · 2007 [cited by examiner]
US 20080285845A1 · Kang · 2008 [cited by examiner]
US 20180262705A1 · Park · 2018 [cited by examiner]
US 20200029033A1 · Xu et al. · 2020 [cited by applicant]
US 20200410272A1 · Seo et al. · 2020 [cited by applicant]
US 20210044769A1 · Maruyama · 2021 [cited by examiner]
US 20220053154A1 · Kitano · 2022 [cited by applicant]
US 20230209220A1 · Nishio · 2023 [cited by examiner]
JP 2022121991A · 2022 [cited by applicant]
WO 2022152619A1 · 2022 [cited by applicant]
Yahiaoui et al., “Overview and Empirical Analysis of ISP Parameter Tuning for Visual Perception in Autonomous Driving”, Journal of Imaging, Sep. 24, 2019, 27 total pages, vol. 5, No. 78, doi: 10.3390/jimaging5100078. [cited by applicant]