IP Library Granted Patent US 12705702
Granted Patent B2
US 12705702 · App. 18/849,552 · Granted Aug 11, 2026

Improving vision pipeline robustness with metadata

Inventors: Yupeng Zhao (Stuttgart, DE); Kensei Jo (Stuttgart, DE); Andreas Aumiller (Stuttgart, DE)
Assignee: SONY SEMICONDUCTOR SOLUTIONS CORPORATION
G06T5/50G06T2207/20221
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 12705702
App. No.
18/849,552
Filed
Sep 23, 2024
Granted
Aug 11, 2026
Kind
B2
Examiner
CHATLY, AMIT
Art Unit
2624
USPC
345/629
Abstract

A system comprising circuitry configured to perform an image processing task, the circuitry comprising a multi-vision solution configured to provide APS data and EVS data, and the circuitry being configured to generate metadata and to provide the metadata to an image processing pipeline comprising an algorithm configured to perform the image processing task.

Claims (18)

1 . A system comprising circuitry configured to perform an image processing task, the circuitry comprising a multi-vision solution configured to provide active pixel sensor (APS) data and event-based sensor (EVS) data, and the circuitry being configured to generate metadata and to provide the metadata to an image processing pipeline comprising an algorithm configured to perform the image processing task, wherein the metadata comprises information on event pixel location; wherein the metadata comprises a saturated area mask; and wherein the metadata comprises one or more of a white balance value, a tone reproduction gain, a gamma gain, an APS pixel noise sigma, an APS pixel gain, an APS exposure timestamps, an EVS contrast threshold value, IMU data, camera calibration data, sensor temperature, or an intermediate image from inside the image processing pipeline.

2 . The system of claim 1 , wherein the image processing pipeline is configured to use the metadata to counteract changes in an image-event data stream relationship.

3 . The system of claim 1 , wherein the algorithm is optimized to a certain image-event output relationship.

4 . The system of claim 1 , wherein the metadata comprises APS metadata obtained from the multi-vision solution, EVS metadata obtained from the multi-vision solution, camera metadata, and/or extra metadata from additional sensors.

5 . The system of claim 1 , wherein circuitry is configured to provide the metadata to a preprocessing of the image processing pipeline.

6 . The system of claim 1 , wherein the image processing pipeline comprises an event preprocessing, and wherein the event pre-processing is configured to use the metadata to pre-process the EVS data or to pre-process data derived from the EVS data.

7 . The system of claim 1 , wherein circuitry is configured to provide the metadata to a post-processing of the image processing pipeline.

8 . The system of claim 1 , wherein circuitry is configured to provide the metadata to a blending, the blending being configured to blend APS data or data derived from the APS data with EVS data or data derived from the EVS data based on the metadata.

9 . The system of claim 1 , wherein the circuitry is configured to provide the metadata directly to the algorithm of the image processing pipeline.

10 . The system of claim 1 , wherein the hybrid sensor comprises circuitry configured to generate the saturated area mask.

11 . The system of claim 1 , wherein the image processing pipeline is configured to generate the saturated area mask.

12 . The system of claim 1 , wherein the image processing pipeline comprises an event transformation.

13 . The system of claim 1 , wherein performing the image processing task comprises generating a target image.

14 . The system of claim 1 , wherein performing the image processing task comprises generating a target image and performing a post-processing.

15 . The system of claim 1 , wherein the multi-vision solution comprises a hybrid vision sensor comprising an EVS sensor and an APS sensor.

16 . The system of claim 1 , wherein the multi-vision solution comprises a multi-camera system.

17 . A method for performing an image processing task, comprising obtaining active pixel sensor (APS) data and event-based sensor (EVS) data from a multi-vision solution, generating metadata, and providing the metadata to an image processing pipeline comprising an algorithm configured to perform the image processing task, wherein the metadata comprises information on event pixel location; wherein the metadata comprises a saturated area mask; and wherein the metadata comprises one or more of a white balance value, a tone reproduction gain, a gamma gain, an APS pixel noise sigma, an APS pixel gain, an APS exposure timestamps, an EVS contrast threshold value, IMU data, camera calibration data, sensor temperature, or an intermediate image from inside the image processing pipeline.

18 . A non-transitory computer readable medium comprising a program comprising instructions, the instructions being configured to, when operated by a processor, perform the method of claim 17 .