Patents by Inventor Xi Peng
Xi Peng has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).
-
Publication number: 20210353262Abstract: A method for positioning key features of a lens based on ocular B-mode ultrasound images includes: acquiring and preprocessing the ocular B-mode ultrasound images to obtain a preprocessed B-mode ultrasound image, eyeball coordinates and lens coordinates; sending the preprocessed B-mode ultrasound image, the eyeball coordinates and the lens coordinates into a trained target detection network YOLOv3 to obtain eyeball position images and lens position images; substituting the eyeball position images and the lens position images into a trained feature extraction network group to obtain image features and feature coordinates corresponding to the eyeball position images and the lens position images, respectively; substituting the image features into a trained collaborative learning network to screen key image features; and marking a feature coordinate corresponding to the key image features on the ocular B-mode ultrasound images to complete positioning the key features of the lens.Type: ApplicationFiled: April 9, 2021Publication date: November 18, 2021Applicant: Sichuan UniversityInventors: Jiancheng Lv, Yong Wang, Yongsheng Sang, Dezhong Peng, Xi Peng, Yanan Sun, Zhenan He, Qing Ye, Mao Li, Quanhui Liu
-
Publication number: 20210274359Abstract: The present disclosure relates to capacity planning methods and apparatus. One example method includes matching a distribution model based on a quantity of service packets in each transmission time interval within specified duration to obtain a matched first distribution model, matching a distribution model based on a length of the service packets to obtain a second distribution model, and performing bandwidth control based on the first distribution model, the second distribution model, a distribution parameter of the first distribution model, and a distribution parameter of the second distribution model.Type: ApplicationFiled: May 14, 2021Publication date: September 2, 2021Inventors: Xi PENG, Bo BAI, Gong ZHANG, Yu LAN, Haofeng QI
-
Publication number: 20210142550Abstract: Aspects of the subject disclosure may include, for example, a method comprising: receiving, by a processing system including a processor, an input three-dimensional dataset comprising a first plurality of two-dimensional images of all or a portion of a subject; applying, by the processing system, bias field correction to the input three-dimensional dataset to generate a corrected three-dimensional dataset comprising a second plurality of two-dimensional images; and generating, by the processing system, a labeled three-dimensional dataset comprising a third plurality of two-dimensional images, wherein the labeled three-dimensional dataset further comprises one or more labels indicating an anatomical structure, and wherein the labeled three-dimensional dataset is generated via a convolutional neural network based upon the corrected three-dimensional dataset and based upon a previously trained three-dimensional dataset. Additional embodiments are disclosed.Type: ApplicationFiled: December 17, 2020Publication date: May 13, 2021Applicant: THE BOARD OF TRUSTEES OF THE UNIVERSITY OF ILLINOISInventors: Bradley P. Sutton, Xi Peng, Matthew Bramlet, Kevin Urbain
-
Patent number: 10891778Abstract: Aspects of the subject disclosure may include, for example, a method comprising: receiving, by a processing system including a processor, an input three-dimensional dataset comprising a first plurality of two-dimensional images of all or a portion of a subject; applying, by the processing system, bias field correction to the input three-dimensional dataset to generate a corrected three-dimensional dataset comprising a second plurality of two-dimensional images; and generating, by the processing system, a labeled three-dimensional dataset comprising a third plurality of two-dimensional images, wherein the labeled three-dimensional dataset further comprises one or more labels indicating an anatomical structure, and wherein the labeled three-dimensional dataset is generated via a convolutional neural network based upon the corrected three-dimensional dataset and based upon a previously trained three-dimensional dataset. Additional embodiments are disclosed.Type: GrantFiled: January 9, 2019Date of Patent: January 12, 2021Assignee: THE BOARD OF TRUSTEES OF THE UNIVERSITY OF ILLINOISInventors: Bradley P. Sutton, Xi Peng, Matthew Bramlet, Kevin Urbain
-
Publication number: 20200213061Abstract: The objective of the present disclosure is to provide a method, apparatus and system of ACK/NACK reporting for Cat-M mechanism. Here, a user equipment feeds back an ACK/NACK message to an eNB on PUSCH based on received downlink data, wherein when computing the number of resource elements occupied by the ACK/NACK message on PUSCH, the number of OFDM symbols in a guard period is ruled out. This may effectively lower the UE's PUSCH data code rate, and meanwhile enhance the eNB's decoding performance on PUSCH.Type: ApplicationFiled: December 15, 2017Publication date: July 2, 2020Applicant: Alcatel LucentInventors: Xi Peng, Qing Cao, Huiping Zheng
-
Method and apparatus for denoising magnetic resonance diffusion tensor, and computer program product
Patent number: 10620287Abstract: The application provides a method, apparatus and computer program product for denoising a magnetic resonance diffusion tensor, wherein the method comprises: collecting data of K space; calculating a maximum likelihood estimator of a diffusion tensor according to the collected data of K space; calculating a maximum posterior probability estimator of the diffusion tensor by using sparsity of the diffusion tensor and sparsity of a diffusion parameter and taking the calculating maximum likelihood estimator as an initial value; and calculating the diffusion parameter according to the calculated maximum posterior probability estimator. The application solves the technical problem in the prior art of how to realize high precision denoising of diffusion tensor while not increasing scanning time and affecting spatial resolution, achieves the technical effects of effectively suppressing noises in the diffusion tensor and improving the estimation accuracy of the diffusion tensor.Type: GrantFiled: June 9, 2017Date of Patent: April 14, 2020Assignee: SHENZHEN INSTITUTES OF ADVANCED TECHNOLOGY CHINESE ACADEMY OF SCIENCESInventors: Xi Peng, Dong Liang, Xin Liu, Hairong Zheng -
Publication number: 20200084611Abstract: A method and system for filtering an access point of an operator are presetting parameters of access points for a plurality of operators, wherein the parameters of the access points each comprise a Mobile Country Code (MCC), a Mobile Network Code (MNC) and an operator name; recognizing SIM cards and filtering a matched operator name list from the parameters of the access points based on the Mobile Country Codes and the Mobile Network Codes of the SIM cards; and selecting an operator name from the operator name list, reserving the selected operator name, and, based on a selected operator name and a corresponding SIM card, filtering an access point of the corresponding SIM card from the parameters of the access points.Type: ApplicationFiled: June 30, 2017Publication date: March 12, 2020Inventors: Yan HUANG, Xi PENG
-
Patent number: 10474883Abstract: A computer-implemented method, system, and computer program product is provided for pose-invariant facial recognition. The method includes generating, by a processor using a recognition neural network, a rich feature embedding for identity information and non-identity information for each of one or more images. The method also includes generating, by the processor using a Siamese reconstruction network, one or more pose-invariant features by employing the rich feature embedding for identity information and non-identity information. The method additionally includes identifying, by the processor, a user by employing the one or more pose-invariant features. The method further includes controlling an operation of a processor-based machine to change a state of the processor-based machine, responsive to the identified user in the one or more images.Type: GrantFiled: November 3, 2017Date of Patent: November 12, 2019Assignee: NEC CorporationInventors: Xiang Yu, Kihyuk Sohn, Manmohan Chandraker, Xi Peng
-
Publication number: 20190213779Abstract: Aspects of the subject disclosure may include, for example, a method comprising: receiving, by a processing system including a processor, an input three-dimensional dataset comprising a first plurality of two-dimensional images of all or a portion of a subject; applying, by the processing system, bias field correction to the input three-dimensional dataset to generate a corrected three-dimensional dataset comprising a second plurality of two-dimensional images; and generating, by the processing system, a labeled three-dimensional dataset comprising a third plurality of two-dimensional images, wherein the labeled three-dimensional dataset further comprises one or more labels indicating an anatomical structure, and wherein the labeled three-dimensional dataset is generated via a convolutional neural network based upon the corrected three-dimensional dataset and based upon a previously trained three-dimensional dataset. Additional embodiments are disclosed.Type: ApplicationFiled: January 9, 2019Publication date: July 11, 2019Applicant: THE BOARD OF TRUSTEES OF THE UNIVERSITY OF ILLINOISInventors: Bradley P. Sutton, Xi Peng, Matthew Bramlet, Kevin Urbain
-
Patent number: 10049307Abstract: Technical solutions are described for training an object-recognition neural network that identifies an object in a computer-readable image. An example method includes assigning a first neural network for determining a visual alignment model of the images for determining a normalized alignment of the object. The method further includes assigning a second neural network for determining a visual representation model of the images for recognizing the object. The method further includes determining the visual alignment model by training the first neural network and determining the visual representation model by training the second neural network independent of the first. The method further includes determining a combined object recognition model by training a combination of the first neural network and the second neural network. The method further includes recognizing the object in the image based on the combined object recognition model by passing the image through each of the neural networks.Type: GrantFiled: April 4, 2016Date of Patent: August 14, 2018Assignee: INTERNATIONAL BUSINESS MACHINES CORPORATIONInventors: Sharathchandra U. Pankanti, Xi Peng, Nalini K. Ratha
-
Patent number: 10018698Abstract: Disclosed is a magnetic resonance rapid parameter imaging method and system. The method comprises: obtaining a target undersampled magnetic resonance signal (S10); obtaining prior information of a parameter model (S20); performing sequence reconstruction of a target image according to the undersampled magnetic resonance signal and the prior information to obtain a target image sequence (S30); and substituting the target image sequence into the parameter estimation model to obtain object parameters and to generate parametric images (S40).Type: GrantFiled: December 5, 2014Date of Patent: July 10, 2018Assignee: Shenzhen Institutes Of Advanced Technology Chinese Academy Of SciencesInventors: Xi Peng, Dong Liang, Xin Liu, Hairong Zheng
-
Publication number: 20180129869Abstract: A computer-implemented method, system, and computer program product is provided for pose-invariant facial recognition. The method includes generating, by a processor using a recognition neural network, a rich feature embedding for identity information and non-identity information for each of one or more images. The method also includes generating, by the processor using a Siamese reconstruction network, one or more pose-invariant features by employing the rich feature embedding for identity information and non-identity information. The method additionally includes identifying, by the processor, a user by employing the one or more pose-invariant features. The method further includes controlling an operation of a processor-based machine to change a state of the processor-based machine, responsive to the identified user in the one or more images.Type: ApplicationFiled: November 3, 2017Publication date: May 10, 2018Inventors: Xiang Yu, Kihyuk Sohn, Manmohan Chandraker, Xi Peng
-
METHOD AND APPARATUS FOR DENOISING MAGNETIC RESONANCE DIFFUSION TENSOR, AND COMPUTER PROGRAM PRODUCT
Publication number: 20170363702Abstract: The application provides a method, apparatus and computer program product for denoising a magnetic resonance diffusion tensor, wherein the method comprises: collecting data of K space; calculating a maximum likelihood estimator of a diffusion tensor according to the collected data of K space; calculating a maximum posterior probability estimator of the diffusion tensor by using sparsity of the diffusion tensor and sparsity of a diffusion parameter and taking the calculating maximum likelihood estimator as an initial value; and calculating the diffusion parameter according to the calculated maximum posterior probability estimator. The application solves the technical problem in the prior art of how to realize high precision denoising of diffusion tensor while not increasing scanning time and affecting spatial resolution, achieves the technical effects of effectively suppressing noises in the diffusion tensor and improving the estimation accuracy of the diffusion tensor.Type: ApplicationFiled: June 9, 2017Publication date: December 21, 2017Inventors: Xi PENG, Dong LIANG, Xin LIU, Hairong ZHENG -
Publication number: 20170286809Abstract: Technical solutions are described for training an object-recognition neural network that identifies an object in a computer-readable image. An example method includes assigning a first neural network for determining a visual alignment model of the images for determining a normalized alignment of the object. The method further includes assigning a second neural network for determining a visual representation model of the images for recognizing the object. The method further includes determining the visual alignment model by training the first neural network and determining the visual representation model by training the second neural network independent of the first. The method further includes determining a combined object recognition model by training a combination of the first neural network and the second neural network. The method further includes recognizing the object in the image based on the combined object recognition model by passing the image through each of the neural networks.Type: ApplicationFiled: April 4, 2016Publication date: October 5, 2017Inventors: Sharathchandra U. Pankanti, Xi Peng, Nalini K. Ratha
-
Publication number: 20160370444Abstract: Disclosed is a magnetic resonance rapid parameter imaging method and system. The method comprises: obtaining a target undersampled magnetic resonance signal (S10); obtaining prior information of a parameter model (S20); performing sequence reconstruction of a target image according to the undersampled magnetic resonance signal and the prior information to obtain a target image sequence (S30); and substituting the target image sequence into the parameter estimation model to obtain object parameters and to generate parametric images (S40).Type: ApplicationFiled: December 5, 2014Publication date: December 22, 2016Applicant: SHENZHEN INSTITUTES OF ADVANCED TECHNOLOGY CHINESE ACADEMY OF SCIENCESInventors: Xi PENG, Dong LIANG, Xin LIU, Hairong ZHENG
-
Patent number: 8624939Abstract: A liquid crystal display (LCD) device includes a display control circuit to receive data of each frame of an external image signal and a liquid crystal panel. When the data of an n+1 frame currently received is the same as that of an n frame previously received, the display control circuit outputs first gray scale voltages corresponding to the data of the n+1 frame to drive the liquid crystal panel. When the data of the n+1 frame is different from that of the n frame, the display control circuit generates data of at least one inserted frame between the data of the n and the n+1 frames, and outputs second gray scale voltages respectively corresponding to the data of the at least one inserted frame and the n+1 frame to drive the liquid crystal panel to display first an image of the least one inserted frame and then that of the n+1 frame. An absolute value of a second gray scale voltage exceeds that of a first gray scale voltage for a same gray scale.Type: GrantFiled: June 18, 2010Date of Patent: January 7, 2014Assignees: Innocom Technology (Shenzhen) Co., Ltd., Chimei Innolux CorporationInventors: Huan-Xi Peng, Sha Feng
-
Publication number: 20110254825Abstract: An LCD includes a liquid crystal panel having a plurality of pixel units each including a liquid crystal capacitor, a scanning circuit for providing scanning signals to the pixel units, a data circuit for providing gray-scale voltage signals to the pixel units; and a timing controller for receiving at least one timing signal and providing a timing control signal to control a driving timing of the scanning circuit and the data circuit according to at least one timing signal. The timing controller outputs a reset control signal to the scanning circuit upon detecting that the liquid crystal display enters a power-off state based on the at least one timing signal. The reset control signal directs the scanning signal to activate all the pixel units to discharge liquid crystal capacitors. A method for driving a liquid crystal display is also provided.Type: ApplicationFiled: December 8, 2010Publication date: October 20, 2011Applicants: CHIMEI INNOLUX CORPORATION, INNOCOM TECHNOLOGY (SHENZHEN) CO., LTD.Inventors: HUAN-XI PENG, SHA FENG
-
Publication number: 20110122166Abstract: A liquid crystal display (LCD) device includes a display control circuit to receive data of each frame of an external image signal and a liquid crystal panel. When the data of an n+1 frame currently received is the same as that of an n frame previously received, the display control circuit outputs first gray scale voltages corresponding to the data of the n+1 frame to drive the liquid crystal panel. When the data of the n+1 frame is different from that of the n frame, the display control circuit generates data of at least one inserted frame between the data of the n and the n+1 frames, and outputs second gray scale voltages respectively corresponding to the data of the at least one inserted frame and the n+1 frame to drive the liquid crystal panel to display first an image of the least one inserted frame and then that of the n+1 frame. An absolute value of a second gray scale voltage exceeds that of a first gray scale voltage for a same gray scale.Type: ApplicationFiled: June 18, 2010Publication date: May 26, 2011Applicants: INNOCOM TECHNOLOGY (SHENZHEN) CO., LTD., CHIMEI INNOLUX CORPORATIONInventors: HUAN-XI PENG, SHA FENG
-
Publication number: 20100013754Abstract: A liquid crystal display (LCD) device includes data lines, pixel units, a first repair line intersecting and disposed on a first side of the data line, a second repair line intersecting and disposed on a second side of the data line, a delay circuit connected between the first repair line and the second repair line, and a control circuit. The control circuit controls a delay time of the delay circuit according to a position of each pixel unit corresponding to a broken data line.Type: ApplicationFiled: July 20, 2009Publication date: January 21, 2010Inventors: Huan-Xi Peng, Sha Feng
-
Publication number: 20090195657Abstract: An onscreen display (OSD) system for a display device includes an OSD module including a microprocessor and a memory. The memory stores instructions about adjustment of a display characteristic and the microprocessor reads the instructions and generates an OSD menu including the instructions. An operation method for the system is also provided.Type: ApplicationFiled: February 2, 2009Publication date: August 6, 2009Inventors: Huan-Xi Peng, Sha Feng