Patents by Inventor Duan Li Liao
Duan Li Liao 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: 11238624Abstract: Provided are an image transform method and an image transform network. The method is for the image transform network including an image generator, a transform discriminator and a focus discriminator, and includes: generating a transformed image according to an un-transformed image and focus information by the image generator; computing a transform discrimination value according to the transformed image by the transform discriminator; computing a value of a first generator loss function and updating the image generator by the image generator; generating a focus discrimination value according to the un-transformed image, the transformed image, and the focus information by the focus discriminator; and computing a value of a second generator loss function according to the focus discrimination value and updating the image generator according to the value of the second generator loss function by the image generator.Type: GrantFiled: March 3, 2020Date of Patent: February 1, 2022Assignee: Industrial Technology Research InstituteInventors: Duan-Li Liao, Chia-Chang Li, Wen-Hung Ting, Po-Lung Chen
-
Publication number: 20210118195Abstract: Provided are an image transform method and an image transform network. The method is for the image transform network including an image generator, a transform discriminator and a focus discriminator, and includes: generating a transformed image according to an un-transformed image and focus information by the image generator; computing a transform discrimination value according to the transformed image by the transform discriminator; computing a value of a first generator loss function and updating the image generator by the image generator; generating a focus discrimination value according to the un-transformed image, the transformed image, and the focus information by the focus discriminator; and computing a value of a second generator loss function according to the focus discrimination value and updating the image generator according to the value of the second generator loss function by the image generator.Type: ApplicationFiled: March 3, 2020Publication date: April 22, 2021Applicant: Industrial Technology Research InstituteInventors: Duan-Li Liao, Chia-Chang Li, Wen-Hung Ting, Po-Lung Chen
-
Patent number: 10890979Abstract: A controlling system and a controlling method for virtual display are provided. The controlling system for virtual display includes a visual line tracking unit, a space forming unit, a hand information capturing unit, a transforming unit and a controlling unit. The visual line tracking unit is used for tracking a visual line of a user. The space forming unit is used for forming a virtual display space according to the visual line. The hand information capturing unit is used for obtaining a hand location of the user's one hand in a real operation space. The transforming unit is used for transforming the hand location to be a cursor location in the virtual display space. The controlling unit is used for controlling the virtual display according to the cursor location.Type: GrantFiled: April 23, 2018Date of Patent: January 12, 2021Assignee: INDUSTRIAL TECHNOLOGY RESEARCH INSTITUTEInventors: Duan-Li Liao, Chia-Chang Li, Wen-Hung Ting, Hian-Kun Tenn
-
Patent number: 10474893Abstract: An electronic device, an iris recognition method and a non-volatile computer-readable medium are provided. A processor in the electronic device obtains a first and second iris images, and calculates a plurality of first and second feature mark boxes that are non-uniformly arranged according to the first and second iris images. The processor uses the first feature mark boxes to obtain a first and second image features from the first and second iris images respectively, and compares the first and second image features to obtain a first recognition result. The processor uses the second feature mark boxes to obtain a third and fourth image features from the second and first iris images respectively, and compares the third and fourth image features to obtain a second recognition result. The processor determines a similarity degree of the first and second iris images according to the first and second recognition results.Type: GrantFiled: April 24, 2018Date of Patent: November 12, 2019Assignee: Industrial Technology Research InstituteInventors: Wen-Hung Ting, Hian-Kun Tenn, Duan-Li Liao, Chia-Chang Li, Po-Lung Chen
-
Publication number: 20190324550Abstract: A controlling system and a controlling method for virtual display are provided. The controlling system for virtual display includes a visual line tracking unit, a space forming unit, a hand information capturing unit, a transforming unit and a controlling unit. The visual line tracking unit is used for tracking a visual line of a user. The space forming unit is used for forming a virtual display space according to the visual line. The hand information capturing unit is used for obtaining a hand location of the user's one hand in a real operation space. The transforming unit is used for transforming the hand location to be a cursor location in the virtual display space. The controlling unit is used for controlling the virtual display according to the cursor location.Type: ApplicationFiled: April 23, 2018Publication date: October 24, 2019Applicant: INDUSTRIAL TECHNOLOGY RESEARCH INSTITUTEInventors: Duan-Li LIAO, Chia-Chang LI, Wen-Hung TING, Hian-Kun TENN
-
Publication number: 20190303645Abstract: An electronic device, an iris recognition method and a non-volatile computer-readable medium are provided. A processor in the electronic device obtains a first and second iris images, and calculates a plurality of first and second feature mark boxes that are non-uniformly arranged according to the first and second iris images. The processor uses the first feature mark boxes to obtain a first and second image features from the first and second iris images respectively, and compares the first and second image features to obtain a first recognition result. The processor uses the second feature mark boxes to obtain a third and fourth image features from the second and first iris images respectively, and compares the third and fourth image features to obtain a second recognition result. The processor determines a similarity degree of the first and second iris images according to the first and second recognition results.Type: ApplicationFiled: April 24, 2018Publication date: October 3, 2019Applicant: Industrial Technology Research InstituteInventors: Wen-Hung Ting, Hian-Kun Tenn, Duan-Li Liao, Chia-Chang Li, Po-Lung Chen
-
Patent number: 9456219Abstract: An image encoding system and an image encoding method are provided. The image encoding system comprises a relation computing unit and an embedded unit. The relation computing unit receives a plurality of view-angle images each comprising a plurality of pixels, and calculates a relative relationship between pixels according to their colors and locations of the pixels to generate a plurality of encoding images. The embedded unit respectively embeds the encoding images in the view-angle images to generate a plurality of embedded images.Type: GrantFiled: September 23, 2011Date of Patent: September 27, 2016Assignee: INDUSTRIAL TECHNOLOGY RESEARCH INSTITUTEInventors: Duan-Li Liao, Chia-Chang Li, Po-Lung Chen
-
Patent number: 9117138Abstract: According to an exemplary embodiment, a method for object positioning by using depth images is executed by a hardware processor as following: converting depth information of each of a plurality of pixels in each of one or more depth images into a real world coordinate; based on the real world coordinate, computing a distance of each pixel to an edge in each of a plurality of directions; assigning a weight to the distance of each pixel to each edge; and based on the weight of the distance of each pixel to each edge and a weight limit, selecting one or more extremity positions of an object.Type: GrantFiled: December 24, 2012Date of Patent: August 25, 2015Assignee: INDUSTRIAL TECHNOLOGY RESEARCH INSTITUTEInventors: Chia-Chang Li, Duan-Li Liao
-
Patent number: 9036920Abstract: A method of detecting feature points of an object in a system for motion detection includes obtaining a first image of the object from a first camera and a second image of the object from a second camera, extracting a foreground image from each of the first image and the second image, based on an assumption that the foreground image is a T-pose image, segmenting the foreground image into a first set of sections, identifying a first set of feature points associated with the first set of sections, obtaining a T-pose image with a set of predetermined feature points, and determining whether the foreground image is a T-pose image by comparing the first set of feature points with the set of predetermined feature points.Type: GrantFiled: October 8, 2010Date of Patent: May 19, 2015Assignees: INDUSTRIAL TECHNOLOGY RESEARCH INSTITUTE, CYWEE GROUP LTD.Inventors: Po-Lung Chen, Chien-Chun Kuo, Wen-Yang Wang, Duan-Li Liao
-
Patent number: 8659547Abstract: A method for constructing a gesture mouse utilizes an object e.g., a user's palm) in the background of the photographed region of a video camera as a determination basis. In a computer system, (1) for the image of an object, the point with maximum Y-axis value is set to be a cursor; (2) the point with maximum X-axis value is set to be a push button; (3) the cursor and the push button are used as centers to set small tracking areas respectively; (4) if the distance between the cursor and the push button becomes greater than a designated threshold within a time period, an action of the push button is determined to occur. Images obtained by the video camera are input into a computer through the digital image interface of the video camera or an analog/digital converting card for further processing.Type: GrantFiled: October 20, 2009Date of Patent: February 25, 2014Assignee: Industrial Technology Research InstituteInventors: Chen Lan Yen, Wen Hung Ting, Chia Chang Li, Duan Li Liao
-
Patent number: 8648866Abstract: A facial animation production method for producing 3-dimensional (3D) facial animation data in response to input video data includes the following steps. First, data positioning and character sorting processes are performed on the input video data to acquire first-layer character data, for indicating multiple first-layer character points, and first-layer model data. Next, first-layer model outline data and first-layer character outline data are respectively obtained according to the first-layer model data and the first-layer character data. Then, the first-layer character outline data is compared with the first-layer model outline data to judge whether a judgment condition is satisfied. If not, output character data are produced according to the first-layer character data, and fundamental facial-mesh transformation data are thus produced. Thereafter, the 3D facial animation data are displayed according to the fundamental facial-mesh transformation data.Type: GrantFiled: July 7, 2010Date of Patent: February 11, 2014Assignee: Industrial Technology Research InstituteInventors: Wen-Hung Ting, Chen-Lan Yen, Wen-Liang Chi, Duan-Li Liao
-
Patent number: 8643643Abstract: An image depth information refreshing method applied to an image depth information refreshing device is disclosed. The image depth information refreshing device includes a storage unit and an arithmetic logic unit. The storage unit stores a previous depth diagram corresponding to a previous image. The image depth information refreshing method includes the following steps. The arithmetic logic unit divides a current image captured by an image capturing unit into multiple image blocks. The arithmetic logic unit calculates a depth reference value of each image block according to the previous depth diagram, and determines a weight of each image block according to the depth reference values. The arithmetic logic unit determines a refreshing frequency of each image block according to the weights. The arithmetic logic unit respectively refreshes the image block according to the refreshing frequencies of the image blocks, and generates a current depth diagram corresponding to the current image.Type: GrantFiled: December 30, 2010Date of Patent: February 4, 2014Assignee: Industrial Technology Research InstituteInventors: Chia-Chang Li, Duan-Li Liao, Wen-Hung Ting, Po-Lung Chen
-
Patent number: 8615136Abstract: A computing device for motion detection in a system capable of detecting feature points of an object of interest is disclosed. The computing device includes a vector forming unit to form a plurality of vectors associated with a set of the feature points and form a vector set based on the vectors, a posture identifying unit to identify a match of a posture in a database based on the vector set, a motion similarity unit to identify a set of predetermined postures in the database based on the matched posture and an immediately previous matched posture, and a motion identifying unit to identify a predetermined motion in the database based on the set of predetermined postures.Type: GrantFiled: October 8, 2010Date of Patent: December 24, 2013Assignees: Industrial Technology Research Institute, Cywee Group Ltd.Inventors: Po-Lung Chen, Chien-Chun Kuo, Wen-Yang Wang, Duan-Li Liao
-
Patent number: 8578299Abstract: A computing device in a system for motion detection comprises an image processing device to determine a motion of an object of interest, and a graphical user interface (GUI) module to drive a virtual role based on the motion determined by the image processing device.Type: GrantFiled: October 8, 2010Date of Patent: November 5, 2013Assignees: Industrial Technology Research Institute, Cywee Group Ltd.Inventors: Po-Lung Chen, Chien-Chun Kuo, Wen-Yang Wang, Duan-Li Liao, Zhou Ye, Sheng-Wen Jeng
-
Publication number: 20130028513Abstract: An image encoding system and an image encoding method are provided. The image encoding system comprises a relation computing unit and an embedded unit. The relation computing unit receives a plurality of view-angle images each comprising a plurality of pixels, and calculates a relative relationship between pixels according to their colors and locations of the pixels to generate a plurality of encoding images. The embedded unit respectively embeds the encoding images in the view-angle images to generate a plurality of embedded images.Type: ApplicationFiled: September 23, 2011Publication date: January 31, 2013Applicant: INDUSTRIAL TECHNOLOGY RESEARCH INSTITUTEInventors: Duan-Li Liao, Chia-Chang Li, Po-Lung Chen
-
Publication number: 20120105433Abstract: An image depth information refreshing method applied to an image depth information refreshing device is disclosed. The image depth information refreshing device includes a storage unit and an arithmetic logic unit. The storage unit stores a previous depth diagram corresponding to a previous image. The image depth information refreshing method includes the following steps. The arithmetic logic unit divides a current image captured by an image capturing unit into multiple image blocks. The arithmetic logic unit calculates a depth reference value of each image block according to the previous depth diagram, and determines a weight of each image block according to the depth reference values. The arithmetic logic unit determines a refreshing frequency of each image block according to the weights. The arithmetic logic unit respectively refreshes the image block according to the refreshing frequencies of the image blocks, and generates a current depth diagram corresponding to the current image.Type: ApplicationFiled: December 30, 2010Publication date: May 3, 2012Applicant: INDUSTRIAL TECHNOLOGY RESEARCH INSTITUTEInventors: Chia-Chang Li, Duan-Li Liao, Wen-Hung Ting, Po-Lung Chen
-
Publication number: 20120087540Abstract: A computing device for motion detection in a system capable of detecting feature points of an object of interest is disclosed. The computing device includes a vector forming unit to form a plurality of vectors associated with a set of the feature points and form a vector set based on the vectors, a posture identifying unit to identify a match of a posture in a database based on the vector set, a motion similarity unit to identify a set of predetermined postures in the database based on the matched posture and an immediately previous matched posture, and a motion identifying unit to identify a predetermined motion in the database based on the set of predetermined postures.Type: ApplicationFiled: October 8, 2010Publication date: April 12, 2012Inventors: PO-LUNG CHEN, Chien-Chun Kuo, Wen-Yang Wang, Duan-Li Liao
-
Publication number: 20120089949Abstract: A computing device in a system for motion detection comprises an image processing device to determine a motion of an object of interest, and a graphical user interface (GUI) module to drive a virtual role based on the motion determined by the image processing device.Type: ApplicationFiled: October 8, 2010Publication date: April 12, 2012Inventors: Po-Lung Chen, Chien-Chun Kuo, Wen-Yang Wang, Duan-Li Liao
-
Publication number: 20120087539Abstract: A method of detecting feature points of an object in a system for motion detection includes obtaining a first image of the object from a first camera and a second image of the object from a second camera, extracting a foreground image from each of the first image and the second image, based on an assumption that the foreground image is a T-pose image, segmenting the foreground image into a first set of sections, identifying a first set of feature points associated with the first set of sections, obtaining a T-pose image with a set of predetermined feature points, and determining whether the foreground image is a T-pose image by comparing the first set of feature points with the set of predetermined feature points.Type: ApplicationFiled: October 8, 2010Publication date: April 12, 2012Inventors: PO-LUNG CHEN, Chien-Chun Kuo, Wen-Yang Wang, Duan-Li Liao
-
Publication number: 20110141105Abstract: A facial animation production method for producing 3-dimensional (3D) facial animation data in response to input video data includes the following steps. First, data positioning and character sorting processes are performed on the input video data to acquire first-layer character data, for indicating multiple first-layer character points, and first-layer model data. Next, first-layer model outline data and first-layer character outline data are respectively obtained according to the first-layer model data and the first-layer character data. Then, the first-layer character outline data is compared with the first-layer model outline data to judge whether a judgment condition is satisfied. If not, output character data are produced according to the first-layer character data, and fundamental facial-mesh transformation data are thus produced. Thereafter, the 3D facial animation data are displayed according to the fundamental facial-mesh transformation data.Type: ApplicationFiled: July 7, 2010Publication date: June 16, 2011Applicant: INDUSTRIAL TECHNOLOGY RESEARCH INSTITUTEInventors: Wen-Hung Ting, Chen-Lan Yen, Wen-Liang Chi, Duan-Li Liao