Samsung Electronics Co., Ltd.

Dongwoon Bai

Jungwon Lee

Mojtaba Rahmati

Jung Hyun Bae

Binnan Zhuang

Sungsoo Kim

Yoo Jin Choi

Inyup Kang

Mostafa El-Khamy

Shuangquan Wang

Hyukjoon Kwon

Qingfeng Liu

Federico Penna

Rama Mythili Vadali

Jaewon Yoo

Maiyuran Wijay

Andrea Kang

Kee-Bong Song

Sungyoon Cho

Jooyeol Yang

Heunchul Lee

Shailesh Chaudhari

Zengli Yang

Long Bao

Tae-ui Kim

Junho Lee

Linbo Li

Jaein Kim

Haoyu Ren

Qi Zhan

Jinhong Wu

Farrokh Etezadi

Heejin Roh

Yangwen Liang

Sili Lu

Jang Wook Moon

Tae-Ui Kim

Junghyun Bae

William Mantzel

Hui Won Je

Hanju Kim

Amin Kheradmand

Abhik K Das

Rohit R Ranade

Daeson Kim

Mohamad Monzer Mansour

Yuansheng Cheng

Seongjun Park

A method includes: computing noise data by subtracting, by a processing circuit, a noisy image from a corresponding ground truth image; clustering, by the processing circuit, a plurality of noise values of the noise data based on intensity values of the corresponding ground truth image; permuting, by the processing circuit, a plurality of locations of the noise values of the noise data within each cluster; generating, by the processing circuit, a synthetic noise image based on the permuted locations of the noise values; adding, by the processing circuit, the synthetic noise image to the corresponding ground truth image to generate a synthetic noisy image; and augmenting an image dataset for training a neural network to perform image denoising with the synthetic noisy image.

Systems and methods for image denoising using deep convolutional networks

A method and system include a symbol processing block to generate log likelihood ratios (LLRs) associated with one or more data symbols. The method and system include a channel estimation (CE) module to receive the LLRs from the symbol processing block, and to process iterative CE (ItCE) for new radio (NR) based at least on reference signals and the LLRs. The CE module can process the ItCE with a granularity of one or more resource blocks (RBs) based at least on pilot resource elements (REs) and virtual pilot REs obtained from the LLRs. The CE module can process the ItCE based at least on a frequency domain orthogonal cover codes (FD-OCC) structure of the reference signals. The reference signals can be demodulation reference signals (DMRS) configured in 5G NR. The CE module can process the ItCE by updating a CE result by adding a quantity that represents a contribution obtained from virtual pilot REs.

Iterative channel estimation for new radio (NR)

Provided is a method of decoding, the method including receiving, by a user equipment (UE), a downlink control information (DCI) that is encoded, identifying, by the UE, a first bit position of the DCI as a known bit, and reducing a number of candidate code words for the DCI based on the known bit.

PDCCH decoding with known DCI bits

A method may include determining, for a channel, a target component of precoding information, determining at least a part of the precoding information based on the target component and a decompressed component of the precoding information, and sending, from a user equipment, the at least a part of the precoding information. Determining the at least a part of the precoding information may be based on a correlation between the target component and the decompressed component. The at least a part of the precoding information may include a vector of coefficients. The vector of coefficients may include a vector of complex numbers. At least one of the complex numbers may include an amplitude representing a scaling coefficient and a phase representing a rotation coefficient. The target component may include a target matrix. The decompressed component may include a decompressed matrix. The target matrix may include a linear combination coefficients matrix. The precoding information may include a precoding matrix indicator.

Systems, methods, and apparatus for determining precoding information for beamforming

A method and an apparatus are provided in which a vector is generated from a covariance matrix of a received signal. The vector is input to a neural network to obtain an enhanced vector. The enhanced vector is converted into an enhanced matrix. Interference whitening is performed on the received signal using a whitening filter based on the enhanced matrix.

Method and apparatus for matrix enhancement in MIMO receiver with deep learning

A method and system are provided. The method includes receiving an input frame, performing high frequency noise reduction (HFNR) on the received input frame, performing low frequency noise reduction (LFNR) on the received input frame, and obtaining a hybrid denoised clean frame by fusing an output of the performed HFNR and an output of the performed LFNR.

Method and apparatus for adaptive hybrid fusion

A method of performing scene-dependent lens shading correction (SD-LSC) is provided. The method includes collecting scene information from a Bayer thumbnail of an input image; generating a standard red green blue (sRGB) thumbnail by processing the Bayer thumbnail of the input image to simulate white balance (WB) and pre-gamma blocks; determining a representative color channel ratio of the input image based on the scene information and the sRGB thumbnail; determining an ideal grid gain of the input image based on the representative color channel ratio and a grid gain of the input image; merging the ideal grid gain and the grid gain of the input image to generate a new grid gain; and applying the new grid gain to the input image.

Method and apparatus based on scene dependent lens shading correction

A method and system are provided. The method includes obtaining a difference map between a reference frame and a non-reference frame, estimating a noise variance of the obtained difference map, obtaining merging weights based on the estimated noise variance and the obtained different map, and merging the non-reference frame with the reference frame using the obtained merging weights.

Method and apparatus for adaptive multi-frame image fusion

A system and method for configuring an RF network based on machine learning. In some embodiments, the method includes: receiving, by a first neural network, a first state and a first state transition, the first state including: one or more identifiers for available active ports, and a set of available connections between two or more circuit elements, each of the circuit elements being one of: (1) a first circuit type, (2) a second circuit type that operatively connects a circuit element of the first circuit type to one of the available active ports, and (3) the available active ports; and generating, by the first neural network, a first estimated quality value, for the first state transition.

Hardware allocation in RFIC based on machine learning

The disclosure provides a method of providing implicit channel state information (CSI) feedback from a user equipment (UE). The method includes determining a precoding matrix indicator (PMI) selection decision metric, selecting one of a sub-band (SB) linear combination coefficient (LCC) selection method, a wideband (WB) LCC selection method, or a sub-band group (SBG) LCC selection method, based on the determined PMI selection decision metric, determining, using the selected LCC selection method, PMI indices based on sub-bands configured by a base station, and transmitting the determined PMI indices to the base station.

Method and apparatus for selection of linear combination coefficients for precoding in frequency-selective channels

A method for computing a dominant class of a scene includes: receiving an input image of a scene; generating a segmentation map of the input image, the segmentation map being labeled with a plurality of corresponding classes of a plurality of classes; computing a plurality of area ratios based on the segmentation map, each of the area ratios corresponding to a different class of the plurality of classes of the segmentation map; and outputting a detected dominant class of the scene based on a plurality of ranked labels based on the area ratios.

System and method for providing dominant scene classification by semantic segmentation

A method and apparatus are provided. The method includes generating a dataset for real-world super resolution (SR), training a first generative adversarial network (GAN), training a second GAN, and fusing an output of the first GAN and an output of the second GAN.

Systems and method of training networks for real-world super resolution with unknown degradations

A method and system are provided. The method includes determining a difference map between a reference frame and a non-reference frame, determining a local variance of the reference frame, determining a detail power map based on a difference between the determined local variance and the determined difference map, and determining a detail grade map based on the determined detail power map.

Method and apparatus for multi-frame based detail grade map estimation and adaptive multi-frame denoising

An apparatus and method for successive multi-frame image denoising are herein disclosed. The apparatus includes a first subtractor including a first input to receive a frame of the image, a second input to receive a reference frame, and an output; an absolute value function block including an input connected to the output of the first subtractor and an output; a second subtractor including a first input connected to the output of the absolute value function block, a second input for receiving a first predetermined value, and an output; and a maximum value divider function block including an input connected to the output of the second subtractor and an output for outputting filter weights.

Apparatus and method for successive multi-frame image denoising

Growing community of inventors

San Diego, CA, United States of America

Dongwoon Bai