公开(公告)号：US11533464B2

公开(公告)日：2022-12-20

申请号：US17270276

申请日：2019-08-20

Applicant: SAMSUNG ELECTRONICS CO., LTD.

Inventor： Gleb Sergeevich Milyukov ,  Konstantin Viktorovich Kolchin ,  Aleksandr Vladislavovich Simutin ,  Michael Nikolaevich Rychagov ,  Sergey Alexandrovich Turko

IPC: H04N13/156 ,  H04N19/597 ,  H04N13/15 ,  H04N13/161 ,  H04N13/282 ,  G06N3/063 ,  H04N13/00

Abstract: A method of synthesizing intermediate views of a light field includes selecting a configuration of specific input views of a light field collected by a light field acquiring device, specifying coordinates of intermediate views to be synthesized and inputting the specified coordinates to a neural network, and synthesizing intermediate views based on a scene disparity, a selected configuration of the specific input views, and the specified coordinates of the intermediate views, using a neural network.

公开(公告)号：US10832432B2

公开(公告)日：2020-11-10

申请号：US16556874

申请日：2019-08-30

Applicant: Samsung Electronics Co., Ltd.

Inventor： Valery Valerievich Anisimovskiy ,  Andrey Yurievich Shcherbinin ,  Sergey Alexandrovich Turko

IPC: G06T7/80 ,  G06T3/00 ,  H04N13/271 ,  G06T7/593 ,  G06T3/40 ,  H04N13/00

Abstract: A method for training a convolutional neural network to reconstruct an image. The method includes forming a common loss function basing on the left and right images (IL, IR), reconstructed left and right images (I′L, I′R), disparity maps (dL, dR), reconstructed disparity maps (d′L, d′R) for the left and right images (IL, IR) and the auxiliary images (I″L, I″R) and training the neural network based on the formed loss function.

公开(公告)号：US10585286B2

公开(公告)日：2020-03-10

申请号：US16059668

申请日：2018-08-09

Applicant: Samsung Electronics Co., Ltd.

Inventor： Konstantin Viktorovich Kolchin ,  Gleb Sergeevich Milyukov ,  Sergey Alexandrovich Turko ,  Jae-yeol Ryu ,  Mikhail Vyacheslavovich Popov ,  Stanislav Aleksandrovich Shtykov ,  Andrey Yurievich Shcherbinin ,  Chan-yul Kim ,  Myung-ho Kim

IPC: G02B27/01 ,  G09G3/34 ,  G09G3/36 ,  G09G3/00

Abstract: A system and a method for displaying a scene are provided. The system includes a display configured to emit light, a spatial light modulator configured to modulate input light based on a transparency value, and at least one processor configured to acquire adjustment information including transparency of the spatial light modulator and light intensity information of the display from a plurality of pieces of view information corresponding to the scene and adjust an intensity value of the light emitted from the display and the transparency value of the spatial light modulator based on the adjustment information, wherein the plurality of pieces of view information are optical information of the scene, the optical information having been acquired at a plurality of viewpoints.

公开(公告)号：US11410323B2

公开(公告)日：2022-08-09

申请号：US17085081

申请日：2020-10-30

Applicant: Samsung Electronics Co., Ltd.

Inventor： Valery Valerievich Anisimovskiy ,  Andrey Yurievich Shcherbinin ,  Sergey Alexandrovich Turko

IPC: G06T7/593 ,  G06T7/80 ,  G06T3/40 ,  H04N13/271 ,  G06T3/00 ,  H04N13/00

Abstract: A method for training a convolutional neural network to reconstruct an image. The method includes forming a common loss function basing on the left and right images (IL, IR), reconstructed left and right images (I′L, I′R), disparity maps (dL, dR), reconstructed disparity maps (d′L, d′R) for the left and right images (IL, IR) and the auxiliary images (I″L, I″R) and training the neural network based on the formed loss function.

公开(公告)号：US20200074661A1

公开(公告)日：2020-03-05

申请号：US16556874

申请日：2019-08-30

Applicant: Samsung Electronics Co., Ltd.

Inventor： Valery Valerievich Anisimovskiy ,  Andrey Yurievich Shcherbinin ,  Sergey Alexandrovich Turko

IPC: G06T7/593 ,  G06T7/80 ,  G06T3/00 ,  G06T3/40 ,  H04N13/271

Abstract: A method for training a convolutional neural network to reconstruct an image. The method includes forming a common loss function basing on the left and right images (IL, IR), reconstructed left and right images (I′L, I′R), disparity maps (dL, dR), reconstructed disparity maps (d′L, d′R) for the left and right images (IL, IR) and the auxiliary images (I″L, I″R) and training the neural network based on the formed loss function.