Patents by Inventor Vijayakumar Bhagavatula
Vijayakumar Bhagavatula 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: 20210364817Abstract: A display system includes an optical element positioned in the optical path to receive light from the display module and direct the light into a field of view to display a scene. The optical element has a focal length on an optical axis of the optical element that varies in response to a control signal. A control system is programmed to generate the control signal to cause the optical element to cycle through a series of focal lengths at a frequency. Each focal length of the series corresponds to a focal plane of a series of focal planes. The control system is programmed to vary the light field from the display module to display the scene as a sequence of different image frames, the different image frames being synchronized so that each of the different image frames is presented at a corresponding one of the series of focal planes.Type: ApplicationFiled: March 5, 2019Publication date: November 25, 2021Inventors: Vijayakumar Bhagavatula, Aswin Sankaranarayanan, Jen-Hao Chang
-
Patent number: 11113829Abstract: A system for analyzing images includes a processing device including a receiving module configured to receive an image associated with a target domain, and a domain adaptation module configured to characterize one or more features represented in the received image based on a domain adaptation model. The domain adaptation model is generated using a machine learning algorithm to train the domain adaptation model, and the machine learning algorithm is configured to train the domain adaptation model based on one or more source domain images associated with a source domain, one or more previously acquired images associated with the target domain, and acquired characterization data associated with the target domain. The system also includes an output module configured to output the received image with characterization data identifying one or more features characterized by the domain adaptation module.Type: GrantFiled: August 20, 2019Date of Patent: September 7, 2021Assignees: GM GLOBAL TECHNOLOGY OPERATIONS LLC, CARNEGIE MELLON UNIVERSITYInventors: Jinsong Wang, Yang Zou, Guangyu J. Zou, Vijayakumar Bhagavatula
-
Publication number: 20210056718Abstract: A system for analyzing images includes a processing device including a receiving module configured to receive an image associated with a target domain, and a domain adaptation module configured to characterize one or more features represented in the received image based on a domain adaptation model. The domain adaptation model is generated using a machine learning algorithm to train the domain adaptation model, and the machine learning algorithm is configured to train the domain adaptation model based on one or more source domain images associated with a source domain, one or more previously acquired images associated with the target domain, and acquired characterization data associated with the target domain. The system also includes an output module configured to output the received image with characterization data identifying one or more features characterized by the domain adaptation module.Type: ApplicationFiled: August 20, 2019Publication date: February 25, 2021Inventors: Jinsong Wang, Yang Zou, Guangyu J. Zou, Vijayakumar Bhagavatula
-
Publication number: 20190130220Abstract: A vehicle, system and method of navigating a vehicle. The vehicle and system include a digital camera for capturing a target image of a target domain of the vehicle, and a processor. The processor is configured to: determine a target segmentation loss for training the neural network to perform semantic segmentation of a target image in a target domain, determine a value of a pseudo-label of the target image by reducing the target segmentation loss while providing aa supervision of the training over the target domain, perform semantic segmentation on the target image using the trained neural network to segment the target image and classify an object in the target image, and navigate the vehicle based on the classified object in the target image.Type: ApplicationFiled: April 10, 2018Publication date: May 2, 2019Inventors: Yang Zou, Zhiding Yu, Vijayakumar Bhagavatula, Jinsong Wang
-
Publication number: 20190129040Abstract: The present application generally relates to methods and apparatus for inferring map data from GPS measurements. More specifically, the method and apparatus to receive a plurality of measurement data from vehicular GPS measurements and, optionally, additional sensor data. The system is then operative to generate latitude and longitude errors from the covariance matrix and apply this information to a plurality of measurement data in order to determine an actual location in response to the plurality of measurement data.Type: ApplicationFiled: October 27, 2017Publication date: May 2, 2019Inventors: Eric He, Fan Bai, Curtis L. Hay, Vijayakumar Bhagavatula
-
Patent number: 10199106Abstract: A method includes, in one aspect, performing a read operation on a wordline of a memory device, wherein the wordline comprises a plurality of cells that are expected to be in a first state; based on the read operation, identifying one or more of the plurality of cells that are determined to be in a second state that differs from the first state; encoding data using information pertaining to the identified cells to generate a codeword comprising a plurality of bits to be written to the wordline, with at least one of the plurality of bits, which are to be written to at least one of the identified cells, having a value corresponding to the second state; and writing the generated codeword to the wordline.Type: GrantFiled: September 30, 2015Date of Patent: February 5, 2019Assignee: Carnegie Mellon UniversityInventors: Yongjune Kim, Vijayakumar Bhagavatula
-
Publication number: 20170345503Abstract: A method includes, in one aspect, performing a read operation on a wordline of a memory device, wherein the wordline comprises a plurality of cells that are expected to be in a first state; based on the read operation, identifying one or more of the plurality of cells that are determined to be in a second state that differs from the first state; encoding data using information pertaining to the identified cells to generate a codeword comprising a plurality of bits to be written to the wordline, with at least one of the plurality of bits, which are to be written to at least one of the identified cells, having a value corresponding to the second state; and writing the generated codeword to the wordlineType: ApplicationFiled: September 30, 2015Publication date: November 30, 2017Inventors: Yongjune Kim, Vijayakumar Bhagavatula
-
Patent number: 9454697Abstract: A method and system for uniquely identifying a subject based on an iris image. After obtaining the iris image, the method produces a filtered iris image by applying filters to the iris image to enhance discriminative features of the iris image. The method analyzes an intensity value for pixels in the filtered iris image to produce an iris code that uniquely identifies the subject. The method also creates a segmented iris image by detecting an inner and outer boundary for an iris region in the iris image, and remapping pixels in the iris region, represented in a Cartesian coordinate system, to pixels in the segmented iris image, represented in a log-polar coordinate system, by employing a logarithm representation process. The method also creates a one-dimensional iris string from the iris image by unfolding the iris region by employing a spiral sampling method to obtain sample pixels in the iris region, wherein the sample pixels are the one-dimensional iris string.Type: GrantFiled: September 9, 2014Date of Patent: September 27, 2016Assignee: Biometricore, Inc.Inventors: Marios Savvides, Khalid Harun, Vijayakumar Bhagavatula, Sungwon Park, Yung-Hui Li
-
Publication number: 20160019422Abstract: A method and system for uniquely identifying a subject based on an iris image. After obtaining the iris image, the method produces a filtered iris image by applying filters to the iris image to enhance discriminative features of the iris image. The method analyzes an intensity value for pixels in the filtered iris image to produce an iris code that uniquely identifies the subject. The method also creates a segmented iris image by detecting an inner and outer boundary for an iris region in the iris image, and remapping pixels in the iris region, represented in a Cartesian coordinate system, to pixels in the segmented iris image, represented in a log-polar coordinate system, by employing a logarithm representation process. The method also creates a one-dimensional iris string from the iris image by unfolding the iris region by employing a spiral sampling method to obtain sample pixels in the iris region, wherein the sample pixels are the one-dimensional iris string.Type: ApplicationFiled: September 9, 2014Publication date: January 21, 2016Applicant: BIOMETRICORE, INC.Inventors: Marios Savvides, Khalid Harun, Vijayakumar Bhagavatula, Sungwon Park, Yung-Hui Li
-
Patent number: 8948285Abstract: A communications system including a transmitter and a receiver. The transmitter transmits a signal with normal data symbols and at least one dual-use data symbol, where the dual-use data symbol has user data and more data protection than the normal data symbols. The receiver receives the transmitted signal and decodes the dual-use data symbol and uses information from decoding of the dual-use data symbol to improve a channel estimate for the normal data symbols.Type: GrantFiled: February 27, 2013Date of Patent: February 3, 2015Assignees: GM Global Technology Operations LLC, Carnegie Mellon UniversityInventors: Zheng Li, Fan Bai, Vijayakumar Bhagavatula
-
Patent number: 8930800Abstract: A method for receiving and storing a packet of symbols. The method decodes the packet of symbols using a first decoding algorithm, and if the first decoding algorithm fails to correctly decode the packet of symbols, then the method decodes the packet of symbols using a second decoding algorithm. If the second decoding algorithm fails to decode the packet of symbols, then a third decoding algorithm is used. The third decoding algorithm can be sub-packet decoding, where a first sub-packet is part of the packet of symbols. If the first sub-packet is decoded successfully, then the method generates a channel estimate using the properly decoded information, and then uses that channel estimate to decode a subsequent sub-packet using the channel estimate, where the second sub-packet is a set of symbols that are a portion of the packet of symbols.Type: GrantFiled: August 14, 2012Date of Patent: January 6, 2015Assignees: GM Global Technology Operations LLC, Carnegie Mellon UniversityInventors: Zheng Li, Fan Bai, Vijayakumar Bhagavatula
-
Patent number: 8861799Abstract: A method and system for uniquely identifying a subject based on an iris image. After obtaining the iris image, the method produces a filtered iris image by applying filters to the iris image to enhance discriminative features of the iris image. The method analyzes an intensity value for pixels in the filtered iris image to produce an iris code that uniquely identifies the subject. The method also creates a segmented iris image by detecting an inner and outer boundary for an iris region in the iris image, and remapping pixels in the iris region, represented in a Cartesian coordinate system, to pixels in the segmented iris image, represented in a log-polar coordinate system, by employing a logarithm representation process. The method also creates a one-dimensional iris string from the iris image by unfolding the iris region by employing a spiral sampling method to obtain sample pixels in the iris region, wherein the sample pixels are the one-dimensional iris string.Type: GrantFiled: March 18, 2013Date of Patent: October 14, 2014Assignee: Biometricore, Inc.Inventors: Marios Savvides, Khalid Harun, Vijayakumar Bhagavatula, Sungwon Park, Yung-Hui Li
-
Publication number: 20140241444Abstract: A communications system including a transmitter and a receiver. The transmitter transmits a signal with normal data symbols and at least one dual-use data symbol, where the dual-use data symbol has user data and more data protection than the normal data symbols. The receiver receives the transmitted signal and decodes the dual-use data symbol and uses information from decoding of the dual-use data symbol to improve a channel estimate for the normal data symbols.Type: ApplicationFiled: February 27, 2013Publication date: August 28, 2014Applicants: CARNEGIE MELLON UNIVERSITY, GM GLOBAL TECHNOLOGY OPERATIONS LLCInventors: Zheng Li, Fan Bai, Vijayakumar Bhagavatula
-
Patent number: 8711661Abstract: A method performed by a disk drive, comprising: receiving a plurality of signal samples over a channel in the disk drive; executing a forward pattern-dependent noise prediction (PDNP) operation on the plurality of the signal samples; generating, based on execution of the forward PDNP operation, a first detection of recorded data bits in the plurality of received signal samples; executing a backward PDNP operation on the plurality of the received signal samples; generating, based on execution of the backward PDNP operation, a second detection of recorded data bits in the plurality of received signal samples; comparing the first detection to the second detection; identifying, based on comparing, one or more erasures in the received plurality of signal samples; and generating one or more sequences of bits that promote correction of the one or more erasures.Type: GrantFiled: January 18, 2013Date of Patent: April 29, 2014Assignee: Carnegie Mellon UniversityInventors: Yibin Ng, Vijayakumar Bhagavatula, Kui Cai
-
Publication number: 20130236067Abstract: A method and system for uniquely identifying a subject based on an iris image. After obtaining the iris image, the method produces a filtered iris image by applying filters to the iris image to enhance discriminative features of the iris image. The method analyzes an intensity value for pixels in the filtered iris image to produce an iris code that uniquely identifies the subject. The method also creates a segmented iris image by detecting an inner and outer boundary for an iris region in the iris image, and remapping pixels in the iris region, represented in a Cartesian coordinate system, to pixels in the segmented iris image, represented in a log-polar coordinate system, by employing a logarithm representation process. The method also creates a one-dimensional iris string from the iris image by unfolding the iris region by employing a spiral sampling method to obtain sample pixels in the iris region, wherein the sample pixels are the one-dimensional iris string.Type: ApplicationFiled: March 18, 2013Publication date: September 12, 2013Applicant: Biometricore, Inc.Inventors: Marios Savvides, Khalid Harun, Vijayakumar Bhagavatula, Sungwon Park, Yung-Hui Li
-
Publication number: 20130188463Abstract: A method performed by a disk drive, comprising: receiving a plurality of signal samples over a channel in the disk drive; executing a forward pattern-dependent noise prediction (PDNP) operation on the plurality of the signal samples; generating, based on execution of the forward PDNP operation, a first detection of recorded data bits in the plurality of received signal samples; executing a backward PDNP operation on the plurality of the received signal samples; generating, based on execution of the backward PDNP operation, a second detection of recorded data bits in the plurality of received signal samples; comparing the first detection to the second detection; identifying, based on comparing, one or more erasures in the received plurality of signal samples; and generating one or more sequences of bits that promote correction of the one or more erasures.Type: ApplicationFiled: January 18, 2013Publication date: July 25, 2013Inventors: Yibin Ng, Vijayakumar Bhagavatula, Cai Kui
-
Patent number: 8411910Abstract: A method and system for uniquely identifying a subject based on an iris image. After obtaining the iris image, the method produces a filtered iris image by applying filters to the iris image to enhance discriminative features of the iris image. The method analyzes an intensity value for pixels in the filtered iris image to produce an iris code that uniquely identifies the subject. The method also creates a segmented iris image by detecting an inner and outer boundary for an iris region in the iris image, and remapping pixels in the iris region, represented in a Cartesian coordinate system, to pixels in the segmented iris image, represented in a log-polar coordinate system, by employing a logarithm representation process. The method also creates a one-dimensional iris string from the iris image by unfolding the iris region by employing a spiral sampling method to obtain sample pixels in the iris region, wherein the sample pixels are the one-dimensional iris string.Type: GrantFiled: April 17, 2009Date of Patent: April 2, 2013Assignee: Biometricore, Inc.Inventors: Marios Savvides, Khalid Harun, Vijayakumar Bhagavatula, Sung Won Park, Yunghui Li
-
Patent number: 8345100Abstract: A method is provided for removing an illumination generated shadow in a captured image. Each pixel of the captured input image is plotted on a two dimensional logarithmic graph. A linear axis for the plurality of color sets is determined that is substantially orthogonal to a respective illumination direction of each respective color set. A log-chromaticity value of each plotted pixel is projected on the axis. An orientation of the linear axis is selected to minimize an illumination effect and provide optimum separation between each of the respective color sets on the linear axis. Edges in the input image and illumination invariant image domain are identified. The identified edges of the input image are compared to identify edges in the illumination invariant image domain. A determination is made whether a shadow edge is present in response to the comparison. A shadow-reduced image is generated for scene analysis by a vehicle vision-based system.Type: GrantFiled: July 6, 2010Date of Patent: January 1, 2013Assignee: GM Global Technology Operations LLCInventors: Wende Zhang, Qi Wu, Vijayakumar Bhagavatula
-
Patent number: 8294794Abstract: A method for is provided for creating a shadow-reduced image from a captured image for distinguishing a clear path of travel. Each pixel of a captured input image is plotted according to a two dimensional logarithmic graph. A specific color set relating to an associated color value of a clear path. A linear illumination-invariant axis is determined as a function of the specific color set. An illumination direction for the linear illumination-invariant axis is determined. A log-chromaticity value of each plotted pixel of the specific color set is projected on the axis. Edges in the input image and the illumination-invariant image domain are identified. The identified edges of the input image are compared to identify edges in the illumination-invariant image domain. A determination is made whether a shadow edge is present in response to comparing the edges. A shadow-reduced image is generated for scene analysis by a vehicle vision-based system.Type: GrantFiled: July 6, 2010Date of Patent: October 23, 2012Assignee: GM Global Technology Operations LLCInventors: Wende Zhang, Qi Wu, Vijayakumar Bhagavatula
-
Publication number: 20120008019Abstract: A method is provided for removing an illumination generated shadow in a captured image. Each pixel of the captured input image is plotted on a two dimensional logarithmic graph. A linear axis for the plurality of color sets is determined that is substantially orthogonal to a respective illumination direction of each respective color set. A log-chromaticity value of each plotted pixel is projected on the axis. An orientation of the linear axis is selected to minimize an illumination effect and provide optimum separation between each of the respective color sets on the linear axis. Edges in the input image and illumination invariant image domain are identified. The identified edges of the input image are compared to identify edges in the illumination invariant image domain. A determination is made whether a shadow edge is present in response to the comparison. A shadow-reduced image is generated for scene analysis by a vehicle vision-based system.Type: ApplicationFiled: July 6, 2010Publication date: January 12, 2012Applicant: GM GLOBAL TECHNOLOGY OPERATIONS, INC.Inventors: Wende Zhang, Qi Wu, Vijayakumar Bhagavatula