IP Library Granted Patent US 12,736,804
Granted Patent B2
US 12,736,804 · App. 18/826,466 · Granted Sep 15, 2026

Method for optimizing or online optimizing extrinsic parameters of fisheye-lens cameras applied to surround-view stitching

Inventors: Fan Dong (Beijing, CN); Cho-Han Wu (New Taipei City, TW)
Assignee: VIA TECHNOLOGIES, INC.
G02B27/0012G01M11/0257G06T7/80G06T2207/20221G06T2207/30256
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 12,736,804
App. No.
18/826,466
Granted
Sep 15, 2026
Kind
B2
Abstract

A method and system is provided in a simulation platform for optimizing the extrinsic parameters of fisheye-lens cameras installed on a vehicle. A simulated vehicle is established according to vehicular characteristics of associated actual vehicle over the simulated platform, thereby a lot of simulated checkerboard calibration plates are placed surrounding the simulated vehicle. A lot of simulated fisheye-lens cameras are generated and mounted on the simulated vehicle based on intrinsic parameters associated with actual fisheye lenses, and fisheye images are derived by using the simulated fisheye-lens cameras, respectively. The initially extrinsic parameters of each of the simulated fisheye-lens cameras are calculated over the simulated platform by using the first characteristic points of its own first fisheye image. Those initially extrinsic parameters may be used to optimize stitching images of the vehicle-surrounding-view systems, or be used to calculate installation tolerances when disposing fisheye-lens cameras physically.

Claims (54)

1 . A method of a simulated platform for online optimizing extrinsic parameters of fisheye-lens cameras in stitching surround-view images, the method comprising:

establishing a simulated vehicle in the simulation platform according to vehicular characteristics associated with an actual vehicle;

generating a plurality of simulated fisheye-lens cameras according to intrinsic parameters of simulated fisheye lenses of the plurality of the simulated fisheye-lens cameras;

disposing the plurality of simulated fisheye-lens cameras on the simulated vehicle;

placing a plurality of simulated calibration plates surrounding the simulated vehicle;

obtaining initial extrinsic parameters of each of the simulated fisheye-lens cameras, comprising the steps of:

generating a first fisheye image by each of the simulated fisheye-lens cameras, wherein the first fisheye image comprises an image of one of the plurality of the simulated calibration plates; and

generating the initial extrinsic parameters of each of the simulated fisheye-lens cameras by using a first characteristic point within the image of one of the plurality of the simulated calibration plates;

optimizing the extrinsic parameters of each of the simulated fisheye-lens cameras, comprising the steps of:

generating at least one second fisheye image according to the initial extrinsic parameters of each of the simulated fisheye-lens cameras;

generating a second stitching image by using the second fisheye images of two adjacently simulated fisheye-lens cameras according to a predetermined direction; and

determining if the extrinsic parameters of each of the simulated fisheye-lens cameras is optimized according to errors of overlap regions of the second stitching image.

2 . The method of claim 1 , further comprising the simulated platform automatically places a plurality of predetermined simulated calibration plates surrounding the simulated vehicle when a first shortcut key is activated.

3 . The method of claim 1 , further comprising the simulated platform automatically disposes the plurality of simulated fisheye-lens cameras on the simulated vehicle when a second shortcut key is activated.

4 . The method of claim 3 , further comprising the simulated platform automatically stores the first fisheye images after the first fisheye images are generated.

5 . The method of claim 3 , wherein the simulated fisheye-lens cameras are indicated by a user over the simulated platform.

6 . The method of claim 1 , wherein the simulated platform automatically aligns camera coordinate systems of the simulated fisheye-lens cameras with a world coordinate system after the initial extrinsic parameters of the simulated fisheye-lens cameras are generated.

7 . The method of claim 6 , wherein the simulated platform adjusts a transform matrix and a rotation matrix associated with the initial extrinsic parameters to align the camera coordinate systems with the world coordinate system.

8 . The method of claim 6 , wherein the step of optimizing the extrinsic parameters comprises:

transforming a second characteristic point to a third characteristic point after the camera coordinate systems align with the world coordinate system; and

optimizing the extrinsic parameters of each of the simulated fisheye-lens cameras by determining if the errors of overlap regions of the second stitching image is larger than a stitching-error threshold by referring to the third characteristic point.

9 . The method of claim 8 , further comprising:

adjusting the initial extrinsic parameters of the two adjacently simulated fisheye-lens cameras to generate an updated second stitching image if the errors of overlap regions of the second stitching image is determined larger than the stitching-error threshold; and

determining errors of overlap regions of the updated second stitching image is determined larger than the stitching-error threshold.

10 . The method of claim 8 , wherein the second characteristic point is an edge of the simulated calibration plates, interior vertex point, a traffic line, a texture of the overlap regions recognizable by the simulated platform.

11 . The method of claim 1 , wherein the first characteristic point is an interior vertex point of one of the simulated checkerboard calibration plates.

12 . The method of claim 11 , further comprising:

obtaining a position of the interior vertex point in the world coordinate system by using a calibration-plate-corner detection algorithm; and

generating the initial extrinsic parameters according to the position of the interior vertex point in the world coordinate system.

13 . The method of claim 1 , wherein the first fisheye image is identical to the second fisheye image.

14 . The method of claim 1 , further comprising optimizing the second stitching images after all of the initial extrinsic parameters of the simulated fisheye-lens camera are optimized.

15 . The method of claim 14 , wherein the step of optimizing the second stitching images comprises re-projection-error minimizations, color/brightness adjustments, and smooth transitions to remove jagged edges or inconsistences of the seam regions.

16 . The method of claim 1 , further comprising:

generating a reference extrinsic parameter for one of the adjacently simulated fisheye-lens cameras;

generating a third fisheye image by one of the adjacently simulated fisheye-lens cameras being introduced the reference extrinsic parameters;

generating a third stitching image by using the third fisheye image and the fisheye image generated by the other one of the adjacently simulated fisheye-lens cameras; and

outputting an offset of the reference extrinsic parameters as a tolerance if errors of overlap regions of the third stitching image is determined larger than a stitching-error threshold.

17 . The method of claim 16 , wherein the reference extrinsic parameters are generated by introducing the offset into the initial extrinsic parameters.

18 . The method of claim 16 , further comprising the step of increasing the offset an increment and introducing to the initial extrinsic parameters of the one of the adjacently simulated fisheye-lens cameras if errors of overlap regions of the third stitching image is determined smaller than a stitching-error threshold.

19 . The method of claim 1 , wherein the simulation platform comprises Carla, PerScan, CarSim, VIRES VTD, PTV Vissim, or TESS NG.

20 . A method of a simulated platform for online optimizing extrinsic parameters of fisheye-lens cameras in stitching surround-view images, the method comprising:

generating at least one fisheye images by each of the simulated fisheye-lens cameras;

generating a stitching image by using the fisheye images of two adjacently simulated fisheye-lens cameras according to a predetermined direction;

determining if the extrinsic parameters of each of the simulated fisheye-lens cameras is optimized according to errors of overlap regions of the stitching image;

transforming a first characteristic point to a second characteristic point after camera coordinate systems of each of the simulated fisheye-lens cameras align with a world coordinate system; and

optimizing the extrinsic parameters of each of the simulated fisheye-lens cameras by determining if the errors of overlap regions of the stitching image is larger than a stitching-error threshold by referring to the third characteristic point.

21 . The method of claim 20 , further comprising:

adjusting current extrinsic parameters of the two adjacently simulated fisheye-lens cameras to generate an updated stitching image if the errors of overlap regions of the stitching image is determined larger than a stitching-error threshold; and

determining errors of overlap regions of the updated stitching image is determined larger than the stitching-error threshold.

22 . The method of claim 21 , wherein the steps of generating the stitching image, determining if the extrinsic parameters of each of the simulated fisheye-lens cameras is optimized, transforming the first characteristic point to the third characteristic point, adjusting the current extrinsic parameters of the two adjacently simulated fisheye-lens cameras is larger than the stitching-error threshold, generating the updated stitching image, and determining errors of overlap regions of the updated stitching image is determined larger than the stitching-error threshold are operated by the simulated platform after the fisheye images are generated.

23 . The method of claim 20 , wherein the first characteristic point is an edge of edges or interior vertex points of a calibration plate, a traffic line, or any identifiable textures of the overlap regions of the stitching image.

24 . The method of claim 20 , further comprising optimizing a surround-view stitching image made by all of the simulated fisheye-lens cameras after all the extrinsic parameters of the simulated fisheye-lens cameras are optimized.

25 . The method of claim 24 , wherein the optimizations of the surround-view stitching image comprise re-projection-error minimizations, color/brightness adjustments, and smooth transitions to remove jagged edges or inconsistences of the seam regions.

26 . The method of claim 20 , wherein each of the images is generated by a fisheye-lens camera installed on a vehicle.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Sep 6, 2024
From: DONG, FAN; WU, CHO-HAN
To: VIA TECHNOLOGIES, INC.
Reel/Frame 068507/0510 →
Priority Claims (2)
CN 202311152370.2 · Sep 7, 2023 · national
CN 202411230045.8 · Sep 3, 2024 · national
Continuity (1)
Related Publication 20250086835A1 · Mar 13, 2025
References Cited (77)
US 9553618B2 · Tian · 2017 [cited by examiner]
US 9723272B2 · Lu · 2017 [cited by examiner]
US 9986173B2 · Wang · 2018 [cited by examiner]
US 10504242B2 · Mou · 2019 [cited by examiner]
US 10582186B1 · Baldwin · 2020 [cited by examiner]
US 10625676B1 · Tsimhoni · 2020 [cited by examiner]
US 10911745B2 · Shivalingappa · 2021 [cited by examiner]
US 11055541B2 · Wallin · 2021 [cited by examiner]
US 11257272B2 · Rowell · 2022 [cited by examiner]
US 11609574B2 · Walters · 2023 [cited by examiner]
US 11693417B2 · George · 2023 [cited by examiner]
US 11812153B2 · Liu · 2023 [cited by examiner]
US 11948315B2 · Ren · 2024 [cited by examiner]
US 12026229B2 · Vahedforough · 2024 [cited by examiner]
US 12217458B1 · Yang · 2025 [cited by examiner]
US 12405116B2 · Mu · 2025 [cited by examiner]
US 12511783B1 · Chen · 2025 [cited by examiner]
US 12530799B2 · Wang · 2026 [cited by examiner]
US 20120287232A1 · Natroshvili · 2012 [cited by examiner]
US 20140098229A1 · Lu · 2014 [cited by examiner]
US 20150254825A1 · Zhang · 2015 [cited by applicant]
US 20150329048A1 · Wang · 2015 [cited by examiner]
US 20160088287A1 · Sadi · 2016 [cited by applicant]
US 20180007263A1 · Vandrotti · 2018 [cited by applicant]
US 20180184078A1 · Shivalingappa · 2018 [cited by examiner]
US 20180336297A1 · Sun · 2018 [cited by examiner]
US 20190058870A1 · Rowell · 2019 [cited by applicant]
US 20190102911A1 · Natroshvili · 2019 [cited by examiner]
US 20200150677A1 · Walters · 2020 [cited by examiner]
US 20200293054A1 · George · 2020 [cited by examiner]
US 20200294194A1 · Sun · 2020 [cited by examiner]
US 20200302696A1 · Thompson · 2020 [cited by applicant]
US 20200342652A1 · Rowell · 2020 [cited by examiner]
US 20200388150A1 · Herson · 2020 [cited by applicant]
US 20210082086A1 · Bichu · 2021 [cited by applicant]
US 20210092354A1 · Shivalingappa · 2021 [cited by examiner]
US 20220137636A1 · Yang · 2022 [cited by examiner]
US 20220321859A1 · Kim · 2022 [cited by examiner]
US 20230024474A1 · Ren · 2023 [cited by applicant]
US 20230032613A1 · Duan · 2023 [cited by applicant]
US 20230283906A1 · Liu · 2023 [cited by examiner]
US 20230319218A1 · Ren · 2023 [cited by applicant]
US 20240424988A1 · Burns · 2024 [cited by examiner]
US 20240430574A1 · Burns · 2024 [cited by examiner]
US 20250085192A1 · Dong · 2025 [cited by examiner]
US 20250085530A1 · Dong · 2025 [cited by examiner]
US 20250244571A1 · Dong · 2025 [cited by examiner]
US 20260004459A1 · Wang · 2026 [cited by examiner]
US 20260017877A1 · Sabo · 2026 [cited by examiner]
CN 102356633A · 2012 [cited by applicant]
CN 108765496A · 2018 [cited by applicant]
CN 109407547A · 2019 [cited by applicant]
CN 109615659A · 2019 [cited by applicant]
CN 110677599A · 2020 [cited by applicant]
CN 105765966B · 2020 [cited by applicant]
CN 111698463A · 2020 [cited by applicant]
CN 113093740A · 2021 [cited by applicant]
CN 113191032A · 2021 [cited by applicant]
CN 113496527A · 2021 [cited by applicant]
CN 113905176A · 2022 [cited by applicant]
CN 215972078U · 2022 [cited by applicant]
CN 114339157A · 2022 [cited by applicant]
CN 116051379A · 2023 [cited by applicant]
CN 116485984A · 2023 [cited by applicant]
JP 2020008664A · 2020 [cited by applicant]
KR 102508361B1 · 2023 [cited by applicant]
TW I607911B · 2017 [cited by applicant]
TW 201913561A · 2019 [cited by applicant]
TW 202321678A · 2023 [cited by applicant]
Chinese language office action dated Mar. 12, 2025, issued in application No. TW 113133892. [cited by applicant]
Chinese language office action dated May 12, 2025, issued in application No. TW 113133890. [cited by applicant]
Chinese language office action dated Sep. 12, 2024, issued in application No. TW 112140701. [cited by applicant]
Chinese language office action dated Feb. 23, 2024, issued in application No. TW 112140699. [cited by applicant]
Chinese language office action dated Jan. 15, 2025, issued in application No. TW 113133894. [cited by applicant]
Chinese language office action dated Jan. 15, 2025, issued in application No. TW 113133895. [cited by applicant]
Office Action dated Mar. 16, 2026, issued in application U.S. Appl. No. 18/826,485. [cited by applicant]
Notice of Allowance dated May 27, 2026, issued in application U.S. Appl. No. 18/826,500. [cited by applicant]