Patents by Inventor Alexandru Malaescu
Alexandru Malaescu 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).
-
Patent number: 12249184Abstract: A method to determine activity in a sequence of successively acquired images of a scene, comprises: acquiring the sequence of images; for each image in the sequence of images, forming a feature block of features extracted from the image and determining image specific information including a weighting for the image; normalizing the determined weightings to form a normalized weighting for each image in the sequence of images; for each image in the sequence of images, combining the associated normalized weighting and associated feature block to form a weighted feature block; passing a combination of the weighted feature blocks through a predictive module to determine an activity in the sequence of images; and outputting a result comprising the determined activity in the sequence of images.Type: GrantFiled: August 17, 2023Date of Patent: March 11, 2025Assignee: Tobii Technologies LimitedInventors: Alexandru Malaescu, Dan Filip, Mihai Ciuc, Liviu-Cristian Dutu, Madalin Dumitru-Guzu
-
Publication number: 20230419727Abstract: A method to determine activity in a sequence of successively acquired images of a scene, comprises: acquiring the sequence of images; for each image in the sequence of images, forming a feature block of features extracted from the image and determining image specific information including a weighting for the image; normalizing the determined weightings to form a normalized weighting for each image in the sequence of images; for each image in the sequence of images, combining the associated normalized weighting and associated feature block to form a weighted feature block; passing a combination of the weighted feature blocks through a predictive module to determine an activity in the sequence of images; and outputting a result comprising the determined activity in the sequence of images.Type: ApplicationFiled: August 17, 2023Publication date: December 28, 2023Inventors: Alexandru Malaescu, Dan Filip, Mihai Ciuc, Liviu-Cristian Dutu, Madalin Dumitru-Guzu
-
Patent number: 11776319Abstract: A method to determine activity in a sequence of successively acquired images of a scene, comprises: acquiring the sequence of images; for each image in the sequence of images, forming a feature block of features extracted from the image and determining image specific information including a weighting for the image; normalizing the determined weightings to form a normalized weighting for each image in the sequence of images; for each image in the sequence of images, combining the associated normalized weighting and associated feature block to form a weighted feature block; passing a combination of the weighted feature blocks through a predictive module to determine an activity in the sequence of images; and outputting a result comprising the determined activity in the sequence of images.Type: GrantFiled: July 14, 2020Date of Patent: October 3, 2023Inventors: Alexandru Malaescu, Dan Filip, Mihai Ciuc, Liviu-Cristian Dutu, Madalin Dumitru-Guzu
-
Patent number: 11727273Abstract: The technology relates to tuning a data translation block (DTB) including a generator model and a discriminator model. One or more processors may be configured to receive training data including an image in a second domain. The image in the second domain may be transformed into a first domain with a generator model. The transformed image may be processed to determine one or more outputs with one or more deep neural networks (DNNs) trained to process data in the first domain. An original objective function for the DTB may be updated based on the one or more outputs. The generator and discriminator models may be trained to satisfy the updated objective function.Type: GrantFiled: December 3, 2021Date of Patent: August 15, 2023Inventors: Alexandru Malaescu, Adrian Dorin Capata, Mihai Ciuc, Alina Sultana, Dan Filip, Liviu-Cristian Dutu
-
Publication number: 20220092361Abstract: The technology relates to tuning a data translation block (DTB) including a generator model and a discriminator model. One or more processors may be configured to receive training data including an image in a second domain. The image in the second domain may be transformed into a first domain with a generator model. The transformed image may be processed to determine one or more outputs with one or more deep neural networks (DNNs) trained to process data in the first domain. An original objective function for the DTB may be updated based on the one or more outputs. The generator and discriminator models may be trained to satisfy the updated objective function.Type: ApplicationFiled: December 3, 2021Publication date: March 24, 2022Applicant: FotoNation LimitedInventors: Alexandru Malaescu, Adrian Dorin Capata, Mihai Ciuc, Alina Sultana, Dan Filip, Liviu-Cristian Dutu
-
Patent number: 11244429Abstract: A method of providing a sharpness measure for an image comprises detecting an object region within an image; obtaining meta-data for the image; and scaling the chosen object region to a fixed size. A gradient map is calculated for the scaled object region and compared against a threshold determined for the image to provide a filtered gradient map of values exceeding the threshold. The threshold for the image is a function of at least: a contrast level for the detected object region, a distance to the subject and an ISO/gain used for image acquisition. A sharpness measure for the object region is determined as a function of the filtered gradient map values, the sharpness measure being proportional to the filtered gradient map values.Type: GrantFiled: May 15, 2020Date of Patent: February 8, 2022Assignee: FotoNation LimitedInventors: Florin Nanu, Adrian Bobei, Alexandru Malaescu, Cosmin Clapon
-
Publication number: 20220019776Abstract: A method to determine activity in a sequence of successively acquired images of a scene, comprises: acquiring the sequence of images; for each image in the sequence of images, forming a feature block of features extracted from the image and determining image specific information including a weighting for the image; normalizing the determined weightings to form a normalized weighting for each image in the sequence of images; for each image in the sequence of images, combining the associated normalized weighting and associated feature block to form a weighted feature block; passing a combination of the weighted feature blocks through a predictive module to determine an activity in the sequence of images; and outputting a result comprising the determined activity in the sequence of images.Type: ApplicationFiled: July 14, 2020Publication date: January 20, 2022Applicant: FotoNation LimitedInventors: Alexandru MALAESCU, Dan FILIP, Mihai CIUC, Liviu-Cristian DUTU, Madalin DUMITRU-GUZU
-
Patent number: 11195056Abstract: The technology relates to tuning a data translation block (DTB) including a generator model and a discriminator model. One or more processors may be configured to receive training data including an image in a second domain. The image in the second domain may be transformed into a first domain with a generator model. The transformed image may be processed to determine one or more outputs with one or more deep neural networks (DNNs) trained to process data in the first domain. An original objective function for the DTB may be updated based on the one or more outputs. The generator and discriminator models may be trained to satisfy the updated objective function.Type: GrantFiled: March 12, 2020Date of Patent: December 7, 2021Inventors: Alexandru Malaescu, Adrian Dorin Capata, Mihai Ciuc, Alina Sultana, Dan Filip, Liviu-Cristian Dutu
-
Publication number: 20210089831Abstract: The technology relates to tuning a data translation block (DTB) including a generator model and a discriminator model. One or more processors may be configured to receive training data including an image in a second domain. The image in the second domain may be transformed into a first domain with a generator model. The transformed image may be processed to determine one or more outputs with one or more deep neural networks (DNNs) trained to process data in the first domain. An original objective function for the DTB may be updated based on the one or more outputs. The generator and discriminator models may be trained to satisfy the updated objective function.Type: ApplicationFiled: March 12, 2020Publication date: March 25, 2021Applicant: FotoNation LimitedInventors: Alexandru MALAESCU, Adrian Dorin CAPATA, Mihai CIUC, Alina SULTANA, Dan FILIP, Liviu-Cristian DUTU
-
Publication number: 20200294206Abstract: A method of providing a sharpness measure for an image comprises detecting an object region within an image; obtaining meta-data for the image; and scaling the chosen object region to a fixed size. A gradient map is calculated for the scaled object region and compared against a threshold determined for the image to provide a filtered gradient map of values exceeding the threshold. The threshold for the image is a function of at least: a contrast level for the detected object region, a distance to the subject and an ISO/gain used for image acquisition. A sharpness measure for the object region is determined as a function of the filtered gradient map values, the sharpness measure being proportional to the filtered gradient map values.Type: ApplicationFiled: May 15, 2020Publication date: September 17, 2020Applicant: FotoNation LimitedInventors: Florin NANU, Adrian BOBEI, Alexandru MALAESCU, Cosmin CLAPON
-
Patent number: 10657628Abstract: A method of providing a sharpness measure for an image comprises detecting an object region within an image; obtaining meta-data for the image; and scaling the chosen object region to a fixed size. A gradient map is calculated for the scaled object region and compared against a threshold determined for the image to provide a filtered gradient map of values exceeding the threshold. The threshold for the image is a function of at least: a contrast level for the detected object region, a distance to the subject and an ISO/gain used for image acquisition. A sharpness measure for the object region is determined as a function of the filtered gradient map values, the sharpness measure being proportional to the filtered gradient map values.Type: GrantFiled: May 30, 2019Date of Patent: May 19, 2020Assignee: FotoNation LimitedInventors: Florin Nanu, Adrian Bobei, Alexandru Malaescu, Cosmin Clapon
-
Patent number: 10560690Abstract: A method for dynamically calibrating an image capture device comprises: a) determining a distance (DCRT, DEST) to an object within a scene; b) determining a first lens actuator setting (DACINIT) for the determined distance; c) determining a second lens actuator setting (DACFOCUS) providing maximum sharpness for the object in a captured image of the scene; and d) storing the determined distance (DCRT, DEST) and the first and second lens actuator settings. These steps are repeated at a second determined distance separated from the first determined distance. A calibration correction (ERRNEARPLP, ERRFARPLP) for stored calibrated lens actuator settings (DACNEARPLP, DACFARPLP) is determined as a function of at least: respective differences between the second lens actuator setting (DACFOCUS) and the first lens actuator setting (DACINIT) for each of the first and second determined distances; and the stored calibrated lens actuator settings are adjusted according to the determined calibration corrections.Type: GrantFiled: December 3, 2018Date of Patent: February 11, 2020Assignee: FotoNation LimitedInventors: Alexandru Malaescu, Florin Nanu
-
Patent number: 10455147Abstract: A method of processing an image comprises: acquiring an image of a scene including an object having a recognizable feature. A lens actuator setting providing a maximum sharpness for a region of the image including the object and a lens displacement corresponding to the lens actuator setting are determined. A distance to the object based on the lens displacement is calculated. A dimension of the feature as a function of the distance to the object, the imaged object size and a focal length of a lens assembly with which the image was acquired, is determined. The determined dimension of the feature is employed instead of an assumed dimension of the feature for subsequent processing of images of the scene including the object.Type: GrantFiled: December 9, 2015Date of Patent: October 22, 2019Assignee: FotoNation LimitedInventors: Alexandru Malaescu, Florin Nanu, Stefan Petrescu, Peter Corcoran, Petronel Bigioi
-
Publication number: 20190279342Abstract: A method of providing a sharpness measure for an image comprises detecting an object region within an image; obtaining meta-data for the image; and scaling the chosen object region to a fixed size. A gradient map is calculated for the scaled object region and compared against a threshold determined for the image to provide a filtered gradient map of values exceeding the threshold. The threshold for the image is a function of at least: a contrast level for the detected object region, a distance to the subject and an ISO/gain used for image acquisition. A sharpness measure for the object region is determined as a function of the filtered gradient map values, the sharpness measure being proportional to the filtered gradient map values.Type: ApplicationFiled: May 30, 2019Publication date: September 12, 2019Applicant: FotoNation LimitedInventors: Florin NANU, Adrian BOBEI, Alexandru MALAESCU, Cosmin CLAPON
-
Patent number: 10334152Abstract: A method for acquiring an image comprises acquiring a first image frame including a region containing a subject at a first focus position; determining a first sharpness of the subject within the first image frame; identifying an imaged subject size within the first image frame; determining a second focus position based on the imaged subject size; acquiring a second image frame at the second focus position; and determining a second sharpness of the subject within the second image frame. A sharpness threshold is determined as a function of image acquisition parameters for the first and/or second image frame. Responsive to the second sharpness not exceeding the first sharpness and the sharpness threshold, camera motion parameters and/or subject motion parameters for the second image frame are determined before performing a focus sweep to determine an optimal focus position for the subject.Type: GrantFiled: December 7, 2017Date of Patent: June 25, 2019Assignee: FotoNation LimitedInventors: Florin Nanu, Alexandru Malaescu, Cosmin Clapon
-
Patent number: 10311554Abstract: A method of providing a sharpness measure for an image comprises detecting an object region within an image; obtaining meta-data for the image; and scaling the chosen object region to a fixed size. A gradient map is calculated for the scaled object region and compared against a threshold determined for the image to provide a filtered gradient map of values exceeding the threshold. The threshold for the image is a function of at least: a contrast level for the detected object region, a distance to the subject and an ISO/gain used for image acquisition. A sharpness measure for the object region is determined as a function of the filtered gradient map values, the sharpness measure being proportional to the filtered gradient map values.Type: GrantFiled: January 16, 2018Date of Patent: June 4, 2019Assignee: FotoNation LimitedInventors: Florin Nanu, Adrian Bobei, Alexandru Malaescu, Cosmin Clapon
-
Publication number: 20190158822Abstract: A method for dynamically calibrating an image capture device comprises: a) determining a distance (DCRT, DEST) to an object within a scene; b) determining a first lens actuator setting (DACINIT) for the determined distance; c) determining a second lens actuator setting (DACFOCUS) providing maximum sharpness for the object in a captured image of the scene; and d) storing the determined distance (DCRT, DEST) and the first and second lens actuator settings. These steps are repeated at a second determined distance separated from the first determined distance. A calibration correction (ERRNEARPLP, ERRFARPLP) for stored calibrated lens actuator settings (DACNEARPLP, DACFARPLP) is determined as a function of at least: respective differences between the second lens actuator setting (DACFOCUS) and the first lens actuator setting (DACINIT) for each of the first and second determined distances; and the stored calibrated lens actuator settings are adjusted according to the determined calibration corrections.Type: ApplicationFiled: December 3, 2018Publication date: May 23, 2019Applicant: FotoNation LimitedInventors: Alexandru MALAESCU, Florin NANU
-
Patent number: 10148943Abstract: A method for acquiring an image comprises acquiring a first image frame including a region containing a subject at a first focus position; determining a first sharpness of the subject within the first image frame; identifying an imaged subject size within the first image frame; determining a second focus position based on the imaged subject size; acquiring a second image frame at the second focus position; and determining a second sharpness of the subject within the second image frame. A sharpness threshold is determined as a function of image acquisition parameters for the first and/or second image frame. Responsive to the second sharpness not exceeding the first sharpness and the sharpness threshold, camera motion parameters and/or subject motion parameters for the second image frame are determined before performing a focus sweep to determine an optimal focus position for the subject.Type: GrantFiled: August 8, 2016Date of Patent: December 4, 2018Assignee: FotoNation LimitedInventors: Florin Nanu, Alexandru Malaescu
-
Patent number: 10148945Abstract: A method for dynamically calibrating an image capture device comprises: a) determining a distance (DCRT, DEST) to an object within a scene; b) determining a first lens actuator setting (DACINIT) for the determined distance; c) determining a second lens actuator setting (DACFOCUS) providing maximum sharpness for the object in a captured image of the scene; and d) storing the determined distance (DCRT, DEST) and the first and second lens actuator settings. These steps are repeated at a second determined distance separated from the first determined distance. A calibration correction (ERRNEARPLP, ERRFARPLP) for stored calibrated lens actuator settings (DACNEARPLP, DACFARPLP) is determined as a function of at least: respective differences between the second lens actuator setting (DACFOCUS) and the first lens actuator setting (DACINIT) for each of the first and second determined distances; and the stored calibrated lens actuator settings are adjusted according to the determined calibration corrections.Type: GrantFiled: May 25, 2017Date of Patent: December 4, 2018Assignee: FotoNation LimitedInventors: Alexandru Malaescu, Florin Nanu
-
Publication number: 20180343444Abstract: A method for dynamically calibrating an image capture device comprises: a) determining a distance (DCRT, DEST) to an object within a scene; b) determining a first lens actuator setting (DACINIT) for the determined distance; c) determining a second lens actuator setting (DACFOCUS) providing maximum sharpness for the object in a captured image of the scene; and d) storing the determined distance (DCRT, DEST) and the first and second lens actuator settings. These steps are repeated at a second determined distance separated from the first determined distance. A calibration correction (ERRNEARPLP, ERRFARPLP) for stored calibrated lens actuator settings (DACNEARPLP, DACFARPLP) is determined as a function of at least: respective differences between the second lens actuator setting (DACFOCUS) and the first lens actuator setting (DACINIT) for each of the first and second determined distances; and the stored calibrated lens actuator settings are adjusted according to the determined calibration corrections.Type: ApplicationFiled: May 25, 2017Publication date: November 29, 2018Inventors: Alexandru MALAESCU, Florin NANU