Samsung electronics co., ltd. (20240119570). MACHINE LEARNING MODEL TRAINING USING SYNTHETIC DATA FOR UNDER-DISPLAY CAMERA (UDC) IMAGE RESTORATION simplified abstract

From WikiPatents
Jump to navigation Jump to search

MACHINE LEARNING MODEL TRAINING USING SYNTHETIC DATA FOR UNDER-DISPLAY CAMERA (UDC) IMAGE RESTORATION

Organization Name

samsung electronics co., ltd.

Inventor(s)

Yibo Xu of Plano TX (US)

Weidi Liu of Houston TX (US)

Hamid R. Sheikh of Allen TX (US)

John Seokjun Lee of Allen TX (US)

MACHINE LEARNING MODEL TRAINING USING SYNTHETIC DATA FOR UNDER-DISPLAY CAMERA (UDC) IMAGE RESTORATION - A simplified explanation of the abstract

This abstract first appeared for US patent application 20240119570 titled 'MACHINE LEARNING MODEL TRAINING USING SYNTHETIC DATA FOR UNDER-DISPLAY CAMERA (UDC) IMAGE RESTORATION

Simplified Explanation

The method described in the abstract involves using a spatially-variant point spread function associated with an under-display camera to generate a synthetic sensor image for training a machine learning model.

  • Identification of a spatially-variant point spread function for an under-display camera
  • Generation of a ground truth image
  • Convolution of the ground truth image based on the point spread function to create a synthetic sensor image
  • Providing the synthetic sensor image and ground truth image as an image pair for machine learning model training

Potential Applications

This technology could be applied in the development of advanced under-display camera systems for smartphones, tablets, and other electronic devices.

Problems Solved

This method helps in accurately simulating images captured by under-display cameras, which can improve the performance and image quality of such cameras.

Benefits

The use of a spatially-variant point spread function and machine learning model training can lead to enhanced image processing capabilities and overall camera performance.

Potential Commercial Applications

The technology could be utilized by manufacturers of electronic devices to enhance the quality of under-display cameras in their products, making them more competitive in the market.

Possible Prior Art

Prior art in the field of computational photography and image processing may include similar methods for simulating camera images and training machine learning models for image enhancement.

Unanswered Questions

How does this method compare to traditional camera calibration techniques?

This method utilizes a spatially-variant point spread function for under-display cameras, which may differ from traditional calibration methods. The effectiveness and efficiency of this approach compared to conventional techniques could be further explored.

What impact could this technology have on the development of future camera technologies?

Understanding how this method can improve the performance of under-display cameras may provide insights into the potential advancements in camera technology and image processing in the future.


Original Abstract Submitted

a method includes identifying, using at least one processing device of an electronic device, a spatially-variant point spread function associated with an under-display camera. the spatially-variant point spread function is based on an optical transmission model and a layout of a display associated with the under-display camera. the method also includes generating, using the at least one processing device, a ground truth image. the method further includes performing, using the at least one processing device, a convolution of the ground truth image based on the spatially-variant point spread function in order to generate a synthetic sensor image. the synthetic sensor image represents a simulated image captured by the under-display camera. in addition, the method includes providing, using the at least one processing device, the synthetic sensor image and the ground truth image as an image pair to train a machine learning model to perform under-display camera point spread function inversion.