Network Learning Techniques (e.g., Back Propagation) Patents (Class 382/157)
-
Publication number: 20150117759Abstract: Provided are a system for search and a method for operating thereof. The system for search includes: a preliminary data analysis part which extracts a variety of attributes through analysis of images being input, analyzes a trend about a category as information requested by a client with the image analysis result, and stores the trend analysis result as metadata; an index part which stores the image analysis result, and structuralizes, organizes and stores the stored metadata in order to easily search the metadata; and a search part which extracts trend information matching a category input by a client, from the index part and provides the trend information in a predetermined format.Type: ApplicationFiled: July 24, 2014Publication date: April 30, 2015Applicant: Samsung Techwin Co., Ltd.Inventors: Dong Jun PARK, Yeon Geol RYU, Hak Chul SHIN, Dong Whan JUNG
-
Publication number: 20150117760Abstract: Systems and methods are disclosed for detecting an object in an image by determining convolutional neural network responses on the image; mapping the responses back to their spatial locations in the image; and constructing features densely extract shift invariant activations of a convolutional neural network to produce dense features for the image.Type: ApplicationFiled: October 17, 2014Publication date: April 30, 2015Applicant: NEC Laboratories America, Inc.Inventors: Xiaoyu Wang, Yuanqing Lin, Will Zou, Miao Sun
-
Patent number: 9002101Abstract: According to an embodiment, a recognition device includes a generation unit to select, plural times, groups each including learning samples from a storage unit, learn a classification metric for classifying the groups selected in each selection, and generate an evaluation metric including the classification metrics; a transformation unit to transform a first feature value of an image including an object into a second feature value using the evaluation metric; a calculation unit to calculate similarities of the object to categories in a table using the second feature value and reference feature values; and a registration unit to register the second feature value as the reference feature value in the table associated with the category of the object and register the first feature value as the learning sample belonging to the category of the object in the storage unit. The generation unit performs the generation again.Type: GrantFiled: March 15, 2012Date of Patent: April 7, 2015Assignee: Kabushiki Kaisha ToshibaInventors: Tomohiro Nakai, Toshimitsu Kaneko, Susumu Kubota, Satoshi Ito, Tatsuo Kozakaya
-
Patent number: 8989492Abstract: A first technique of recognizing content is disclosed, including: determining a first value representative of a pixel content present at a first set of pixels associated with a first distance from a pixel under consideration; determining a second value representative of a pixel content present at a second set of pixels associated with a second distance from the pixel under consideration; and using the first and second values to compute one or more spatial features associated with the pixel under consideration for purposes of content recognition. A second technique of recognizing content is also disclosed, including: determining, for a pixel, a first value representative of a first feature associated with a set of pixels associated with a first direction from the pixel; and determining, for the pixel, a second value representative of a second feature associated with a set of pixels associated with a second direction from the pixel.Type: GrantFiled: June 4, 2012Date of Patent: March 24, 2015Assignee: Apple Inc.Inventors: Jerome R. Bellegarda, Jannes G. A. Dolfing
-
Patent number: 8918352Abstract: Learning processes for a single hidden layer neural network, including linear input units, nonlinear hidden units, and linear output units, calculate the lower-layer network parameter gradients by taking into consideration a solution for the upper-layer network parameters. The upper-layer network parameters are calculated by a closed form formula given the lower-layer network parameters. An accelerated gradient algorithm can be used to update the lower-layer network parameters. A weighted gradient also can be used. With the combination of these techniques, accelerated training with faster convergence, to a point with a lower error rate, can be obtained.Type: GrantFiled: May 23, 2011Date of Patent: December 23, 2014Assignee: Microsoft CorporationInventors: Li Deng, Dong Yu
-
Patent number: 8903166Abstract: This document describes techniques that utilize a learning method to generate a ranking model for use in image search systems. The techniques leverage textual information and visual information simultaneously when generating the ranking model. The tools are further configured to apply the ranking model responsive to receiving an image search query.Type: GrantFiled: January 20, 2010Date of Patent: December 2, 2014Assignee: Microsoft CorporationInventors: Linjun Yang, Bo Geng, Xian-Sheng Hua
-
Patent number: 8897579Abstract: A computer-implemented method of managing information is disclosed. The method can include receiving a message from a mobile device configured to connect to a mobile device network (the message including a digital image taken by the mobile device and including information corresponding to words), determining the words from the digital image information using optical character recognition, indexing the digital image based on the words, and storing the digital image for later retrieval of the digital image based on one or more received search terms.Type: GrantFiled: October 9, 2013Date of Patent: November 25, 2014Assignee: Google Inc.Inventors: Krishnendu Chaudhury, Ashutosh Garg, Prasenjit Phukan, Arvind Saraf
-
Patent number: 8885931Abstract: One or more techniques and/or systems are disclosed for mitigating machine solvable human interactive proofs (HIPs). A classifier is trained over a set of one or more training HIPs that have known characteristics for OCR solvability and HIP solving pattern from actual use. A HIP classification is determined for a HIP (such as from a HIP library used by a HIP generator) using the trained classifier. If the HIP is classified by the trained classifier as a merely human solvable classification, such that it may not be solved by a machine, the HIP can be identified for use in the HIP generation system. Otherwise, the HIP can be altered to (attempt to) be merely human solvable.Type: GrantFiled: January 26, 2011Date of Patent: November 11, 2014Assignee: Microsoft CorporationInventor: Kumar S. Srivastava
-
Patent number: 8873838Abstract: The present invention relates to a method and system for characterizing an image. The characterization may then be used to conduct a search for similar images, for example using a learning system trained using previously characterized images. A face may be identified within the image and a subsection extracted from said image which does not contain said face. At least one fixed size patch is taken from said extracted subsection; and input into said learning network to characterize said image.Type: GrantFiled: March 14, 2013Date of Patent: October 28, 2014Assignee: Google Inc.Inventors: Mustafa Suleyman, Benjamin Kenneth Coppin, Marek Barwinski, Arun Nair, Andrei-Alexandru Rusu, Chia-Yueh Carlton Chu
-
Patent number: 8873844Abstract: Systems and methods for metric learning include iteratively determining feature groups of images based on its derivative norm. Corresponding metrics of the feature groups are learned by gradient descent based on an expected loss. The corresponding metrics are combined to provide an intermediate metric matrix as a sparse representation of the images. A loss function of all metric parameters corresponding to features of the intermediate metric matrix are optimized, using a processor, to learn a final metric matrix. Eigenvalues of the final metric matrix are projected onto a simplex.Type: GrantFiled: November 21, 2012Date of Patent: October 28, 2014Assignee: NEC Laboratories America, Inc.Inventors: Chang Huang, Shenghuo Zhu, Kai Yu
-
Publication number: 20140270488Abstract: The present invention relates to a method and system for characterizing an image. The characterization may then be used to conduct a search for similar images, for example using a learning system trained using previously characterized images. A face may be identified within the image and a subsection extracted from said image which does not contain said face. At least one fixed size patch is taken from said extracted subsection; and input into said learning network to characterize said image.Type: ApplicationFiled: March 14, 2013Publication date: September 18, 2014Applicant: Google Inc.Inventors: Mustafa Suleyman, Benjamin Kenneth Coppin, Marek Barwinski, Arun Nair, Andrei-Alexandru Rusu, Chia-Yueh Carlton Chu
-
Patent number: 8811723Abstract: A user input method and apparatus may comprise a two line object imaging sensor having a primary line scan-sensor providing a primary line scan-sensor output and a secondary line scan-sensor providing a secondary line scan-sensor output, representing pixels in a current primary scan row and a current secondary scan row, and adapted to scan an object; storing for each scan time each current primary line scan-sensor output and each current secondary line scan-sensor output and a correlation unit correlating at least one of the current representations of pixels in a primary line sensor output with stored representations and the current representations of pixels in a secondary line sensor output with stored representations and, the correlation unit providing as an output a motion indicator.Type: GrantFiled: August 20, 2013Date of Patent: August 19, 2014Inventor: Anthony Russo
-
Patent number: 8798345Abstract: A diagnosis processing device is provided in which diagnosis is realizable by a simple arrangement. A diagnosis processing device (1) of the present invention includes: a learning pattern creating section (10a) for creating a learning pattern by sampling data from a learning image in which abnormality information indicating a substantive feature of abnormality of a target is pre-known; a learning processing section (12) for causing a neural network (17) to learn, by using learning patterns; a diagnostic pattern creating section (10b) for creating a diagnostic pattern by sampling data from a diagnostic image in which abnormality information is unknown; a determination processing section (18) for determining a substantive feature of the abnormality of the target indicated in the abnormality information in the diagnostic image, based on an output value outputted, in response to an input of the diagnostic pattern, from a learned neural network (17) which is a neural network subjected to learning.Type: GrantFiled: August 16, 2010Date of Patent: August 5, 2014Assignees: Sharp Kabushiki Kaisha, National University Corporation Tottori UniversityInventors: Takahiro Sasaki, Satoru Kishida, Kentaro Kinoshita
-
Patent number: 8731307Abstract: There is provided an information processing apparatus including: a characteristic amount extracting unit extracting a plurality of characteristic amounts, which are information expressing characteristics of a video, from the video; a labeling unit associating the extracted characteristic amounts with a person or a background; a matching degree judging unit judging a degree of matching between the associated characteristic amounts and the characteristic amounts of at least one other video; a comparing unit comparing the plurality of characteristic amounts of one scene in the video from which the characteristic amounts have been extracted and the plurality of characteristic amounts of one scene in the at least one other video; and a relationship inferring unit inferring a relationship between the one scene in the video and the one scene in the at least one other video based on a comparison result of the comparing unit.Type: GrantFiled: March 2, 2011Date of Patent: May 20, 2014Assignee: Sony CorporationInventor: Akifumi Kashiwagi
-
Patent number: 8718393Abstract: An urban scenes reconstruction method includes: acquiring digital data of a three-dimensional subject, the digital data comprising a 2D photograph and a 3D scan; fusing the 3D scan and the 2D photograph to create a depth-augmented photograph; decomposing the depth-augmented photograph into a plurality of constant-depth layers; detecting repetition patterns of each constant-depth layer; and using the repetitions to enhance the 3D scan to generate a polygon-level 3D reconstruction.Type: GrantFiled: July 28, 2010Date of Patent: May 6, 2014Assignee: Shenzhen Institutes of Advanced Technology Chinese Academy of SciencesInventors: Baoquan Chen, Liangliang Nan, Qian Zheng
-
Patent number: 8705826Abstract: A two-dimensional retinal fundus image of the retinal fundus of an eye is processed by optic disc segmentation (2) followed by cup segmentation 4. Data derived from the optic disc segmentation (i.e. the output of the disc segmentation (2) and/or data derived from the output of the optic disc segmentation step, e.g. by a smoothing operation 3) and data derived from the out-put of the optic cup segmentation (i.e. the output of the cup segmentation (4) and/or data derived from the output of the optic disc segmentation, e.g. by a smoothing operation 5) are fed (6) to an adaptive model which has been trained to generate from such inputs a value indicative of cup-to-disc ratio (CDR) of the eye. The CDR is indicative of glaucoma. Thus, the method can be used to screen patients for glaucoma.Type: GrantFiled: May 14, 2008Date of Patent: April 22, 2014Assignee: Agency for Science, Technology and ResearchInventors: Jiang Liu, Joo Hwee Lim, Wing Kee Wong, Huiqi Li, Tien Yin Wong
-
Patent number: 8693765Abstract: The invention includes a method for recognizing shapes using a preprocessing mechanism that decomposes a source signal into basic components called atoms and a recognition mechanism that is based on the result of the decomposition performed by the preprocessing mechanism. In the method, the preprocessing mechanism includes at least one learning phase culminating in a set of signals called kernels, the kernels being adapted to minimize a cost function representing the capacity of the kernels to correctly reconstruct the signals from the database while guaranteeing a sparse decomposition of the source signal while using a database of signals representative of the source to be processed and a coding phase for decomposing the source signal into atoms, the atoms being generated by shifting of the kernels according to their index, each of the atoms being associated with a decomposition coefficient. The invention also includes a shape recognition system for implementing the method.Type: GrantFiled: August 13, 2009Date of Patent: April 8, 2014Assignee: Commissariat a l'Energie Atomique et aux Energies AlternativesInventors: David Mercier, Anthony Larue
-
Patent number: 8649613Abstract: A classifier training system trains unified classifiers for categorizing videos representing different categories of a category graph. The unified classifiers unify the outputs of a number of separate initial classifiers trained from disparate subsets of a training set of media items. The training process divides the training set into a number of bags, and applies a boosting algorithm to the bags, thus enhancing the accuracy of the unified classifiers.Type: GrantFiled: November 3, 2011Date of Patent: February 11, 2014Assignee: Google Inc.Inventors: Thomas Leung, Yang Song, John Zhang
-
Patent number: 8644624Abstract: Embodiments include a scene classification system and method. In one embodiment, a method includes forming a first plurality of image features from an input image, processing the first plurality of image features in the first scene classifier.Type: GrantFiled: July 28, 2009Date of Patent: February 4, 2014Assignee: Samsung Electronics Co., Ltd.Inventors: Li Tao, Yeong-Taeg Kim
-
Patent number: 8625885Abstract: Systems and methods for automated pattern recognition and object detection. The method can be rapidly developed and improved using a minimal number of algorithms for the data content to fully discriminate details in the data, while reducing the need for human analysis. The system includes a data analysis system that recognizes patterns and detects objects in data without requiring adaptation of the system to a particular application, environment, or data content. The system evaluates the data in its native form independent of the form of presentation or the form of the post-processed data.Type: GrantFiled: February 28, 2011Date of Patent: January 7, 2014Assignee: Intelliscience CorporationInventors: Robert M. Brinson, Jr., Nicholas Levi Middleton, Bryan Glenn Donaldson
-
Patent number: 8615475Abstract: Mitigation of processing artifacts caused by surfaces with high contrast printing or coloring transitions within a system to compare signatures derived from inherent physical surface properties of different articles to authenticate or validate articles and within a system to generate signatures from inherent physical surface properties of different articles.Type: GrantFiled: December 18, 2009Date of Patent: December 24, 2013Assignee: Ingenia Holdings LimitedInventor: Russell Paul Cowburn
-
Patent number: 8577130Abstract: Described herein is a technology for facilitating deformable model-based segmentation of image data. In one implementation, the technology includes receiving training image data (202) and automatically constructing a hierarchical structure (204) based on the training image data. At least one spatially adaptive boundary detector is learned based on a node of the hierarchical structure (206).Type: GrantFiled: March 15, 2010Date of Patent: November 5, 2013Assignee: Siemens Medical Solutions USA, Inc.Inventors: Maneesh Dewan, Yiqiang Zhan, Xiang Sean Zhou, Zhao Yi
-
Patent number: 8577131Abstract: Systems and methods for improving visual object recognition by analyzing query images are disclosed. In one example, a visual object recognition module may determine query images matching objects of a training corpus utilized by the module. Matched query images may be added to the training corpus as training images of a matched object to expand the recognition of the object by the module. In another example, relevant candidate image corpora from a pool of image data may be automatically selected by matching the candidate image corpora against user query images. Selected image corpora may be added to a training corpus to improve recognition coverage. In yet another example, objects unknown to a visual object recognition module may be discovered by clustering query images. Clusters of similar query images may be annotated and added into a training corpus to improve recognition coverage.Type: GrantFiled: July 12, 2011Date of Patent: November 5, 2013Assignee: Google Inc.Inventors: Yuan Li, Hartwig Adam
-
Patent number: 8529446Abstract: In a method for determining a parameter in an automatic study and data management system, data is gathered in a knowledge database, and a parameter is determined based the data gathered in the knowledge database. The data is correlated to at least one of a configuration and implementation of a previous clinical study. The parameter is usable for configuring a future clinical study.Type: GrantFiled: May 31, 2007Date of Patent: September 10, 2013Assignee: Siemens AktiengesellschaftInventors: Markus Schmidt, Siegfried Schneider, Gudrun Zahlmann
-
Patent number: 8515207Abstract: A technique for efficient processing of annotations in images such as panoramic images is herein disclosed. In an embodiment, a first user annotation is received for a feature in a first panorama, and a second user annotation for the same feature is received in a second panorama. The coordinates for the feature can then be generated by computing the intersection between data generated for the first user annotation and for the second user annotations.Type: GrantFiled: May 25, 2007Date of Patent: August 20, 2013Assignee: Google Inc.Inventor: Stephen Chau
-
Patent number: 8494257Abstract: Data set generation and data set presentation for image processing are described. The processing determines a location for each of one or more musical artifacts in the image and identifies a corresponding label for each of the musical artifacts, generating a training file that associates the identified labels and determined locations of the musical artifacts with the image, and presenting the training file to a neural network for training.Type: GrantFiled: February 13, 2009Date of Patent: July 23, 2013Assignee: Museami, Inc.Inventors: Robert Taub, George Tourtellot
-
Publication number: 20130148879Abstract: An information processing apparatus includes a network learning portion that performs learning of an appearance/position recognition network by constraining first to third weights and using a learning image, wherein the appearance/position recognition network has a foreground layer including a position node, a background layer including a background node, and an image layer including a pixel node, and is a neural network in which the position node, the background node, and the pixel node are connected to each other, and wherein the first weight is a connection weight between the position node and the pixel node, the second weight is a connection weight between the position node and the background node, and the third weight is a connection weight between the background node and the pixel node.Type: ApplicationFiled: November 30, 2012Publication date: June 13, 2013Applicant: Sony CorporationInventor: Sony Corporation
-
Patent number: 8422767Abstract: The present invention discloses a system and method of transforming a sample of content data by utilizing known samples in a learning best to best determine coefficients for a linear combination of non-linear filter functions and applying the coefficients to the content data in an operational phase.Type: GrantFiled: April 23, 2008Date of Patent: April 16, 2013Assignee: Gabor LigetiInventor: Gabor Ligeti
-
Patent number: 8391542Abstract: Provided is an iterative method of estimating the pose of a moving PTZ camera. The first step is to use an image registration method on a reference image and a current image to calculate a matrix that estimates the motion of sets of points corresponding to the same object in both images. Information about the absolute camera pose, embedded in the matrix obtained in the first step, is used to simultaneously recalculate both the starting positions in the reference image and the motion estimate. The recalculated starting positions and motion estimate are used to determine the pose of the camera in the current image. The current image is taken as a new reference image, a new current image is selected and the process is repeated in order to determine the pose of the camera in the new current image. The entire process is repeated until the camera stops moving.Type: GrantFiled: April 22, 2008Date of Patent: March 5, 2013Assignee: DVTel, Inc.Inventor: Nir Avrahami
-
Patent number: 8379969Abstract: The invention relates to a method for matching an object model to a three-dimensional point cloud, wherein the point cloud is generated from two images by means of a stereo method and a clustering method is applied to the point cloud in order to identify points belonging to respectively one cluster, wherein model matching is subsequently carried out, with at least one object model being superposed on at least one cluster and an optimum position of the object model with respect to the cluster being determined, and wherein a correction of false assignments of points is carried out by means of the matched object model. A classifier, trained by means of at least one exemplary object, is used to generate an attention map from at least one of the images. A number and/or a location probability of at least one object, which is similar to the exemplary object, is determined in the image using the attention map, and the attention map is taken into account in the clustering method and/or in the model matching.Type: GrantFiled: April 8, 2010Date of Patent: February 19, 2013Assignee: Pilz GmbH & Co. KGInventors: Bjoern Barrois, Lars Krueger, Christian Woehler
-
Patent number: 8345962Abstract: A method and system for training a neural network of a visual recognition computer system, extracts at least one feature of an image or video frame with a feature extractor; approximates the at least one feature of the image or video frame with an auxiliary output provided in the neural network; and measures a feature difference between the extracted at least one feature of the image or video frame and the approximated at least one feature of the image or video frame with an auxiliary error calculator. A joint learner of the method and system adjusts at least one parameter of the neural network to minimize the measured feature difference.Type: GrantFiled: November 25, 2008Date of Patent: January 1, 2013Assignee: NEC Laboratories America, Inc.Inventors: Kai Yu, Wei Xu, Yihong Gong
-
Patent number: 8321937Abstract: An intrusion prevention/detection system filter (IPS filter) performance evaluation is provided. The performance evaluation is performed at both the security center and at the customer sites to derive a base confidence score and local confidence scores. Existence of new vulnerability is disclosed and its attributes are used in the generation of new IPS filter or updates. The generated IPS filter is first tested to determine its base confidence score from test confidence attributes prior to deploying it to a customer site. A deep security manager and deep security agent, at the customer site, collect local confidence attributes that are used for determining the local confidence score. The local confidence score and the base confidence score are aggregated to form a global confidence score. The local and global confidence scores are then compared to deployment thresholds to determine whether the IPS filter should be deployed in prevention or detection mode or sent back to the security center for improvement.Type: GrantFiled: October 22, 2008Date of Patent: November 27, 2012Assignee: Trend Micro IncorporatedInventors: Blake Stanton Sutherland, William G. McGee
-
Patent number: 8300955Abstract: A plurality of images inputted in an image signal input portion are divided into a plurality of regions by an image dividing portion, and a feature value in each of the plurality of regions is calculated by a feature value calculation portion and divided into a plurality of subsets by a subset generation portion. On the other hand, a cluster classifying portion classifies a plurality of clusters generated in a feature space into any one of a plurality of classes on the basis of the feature value and occurrence frequency of the feature value. And a classification criterion calculation portion calculates a criterion of classification for classifying images included in one subset on the basis of a distribution state of the feature value in the feature space of each of the images included in the one subset.Type: GrantFiled: June 24, 2011Date of Patent: October 30, 2012Assignee: Olympus Medical Systems Corp.Inventors: Hirokazu Nishimura, Tetsuo Nonami
-
Patent number: 8290250Abstract: An image-based pattern recognizer and a method and apparatus for making such a pattern recognizer are disclosed. By employing positional coding, the meaning of any feature present in an image can be defined implicitly in space. The pattern recognizer can be a neural network including a plurality of stages of observers. The observers are configured to cooperate to identify the presence of features in the input image and to recognize a pattern in the input image based on the features. Each of the observers includes a plurality of neurons. The input image includes a plurality of units, and each of the observers is configured to generate a separate output set that includes zero or more coordinates of such units.Type: GrantFiled: December 26, 2008Date of Patent: October 16, 2012Assignee: Five Apes, Inc.Inventor: Williams J. F. Paquier
-
Patent number: 8284239Abstract: The invention discloses the asynchronous photography for dual camera apparatus and processing the method for real-time forward vehicle detection. Image is captured by a pair of monochrome camera and stored into a computer. After the video pre-process, the edge information is used to locate the forward vehicle position, and then obtained the disparity from a fast comparison search algorithm by the stereo vision methodology. Proposed algorithm calculation of the invention can conquer the asynchronous exposure problem from dual camera and lower the hardware cost.Type: GrantFiled: July 8, 2009Date of Patent: October 9, 2012Assignee: National Defense UniversityInventors: Chung-Cheng Chiu, Wen-Chung Chen, Meng-Liang Chung
-
Patent number: 8260005Abstract: A system, method and device for grading of meat such as bovine, porcine, sheep, horse or poultry meat among others. The device of this invention is a portable tool, which is approached toward a meat specimen to be analyzed and captures an image. The device then objectively relates the image to meat quality parameters by means of an image analyzing method. The device and method solve, in a practical, fast and satisfactory way, the problem of determining meat quality parameters such as texture, color, and contained intramuscular fat percentage.Type: GrantFiled: March 31, 2008Date of Patent: September 4, 2012Assignee: Universidad De Santiago De ChileInventors: Gerda Roxana Tomic, Renato Salinas, José Rolando Silva, Fernando Osorio, Héctor Barrera
-
Patent number: 8229209Abstract: An image-based pattern recognizer and a method and apparatus for making such a pattern recognizer are disclosed. By employing positional coding, the meaning of any feature present in an image can be defined implicitly in space. The pattern recognizer can be a neural network including a plurality of stages of observers. The observers are configured to cooperate to identify the presence of features in the input image and to recognize a pattern in the input image based on the features. Each of the observers includes a plurality of neurons. The input image includes a plurality of units, and each of the observers is configured to generate a separate output set that includes zero or more coordinates of such units.Type: GrantFiled: December 26, 2008Date of Patent: July 24, 2012Assignee: Five Apes, Inc.Inventor: Williams J. F. Paquier
-
Publication number: 20120183187Abstract: A diagnosis processing device is provided in which diagnosis is realizable by a simple arrangement. A diagnosis processing device (1) of the present invention includes: a learning pattern creating section (10a) for creating a learning pattern by sampling data from a learning image in which abnormality information indicating a substantive feature of abnormality of a target is pre-known; a learning processing section (12) for causing a neural network (17) to learn, by using learning patterns; a diagnostic pattern creating section (10b) for creating a diagnostic pattern by sampling data from a diagnostic image in which abnormality information is unknown; a determination processing section (18) for determining a substantive feature of the abnormality of the target indicated in the abnormality information in the diagnostic image, based on an output value outputted, in response to an input of the diagnostic pattern, from a learned neural network (17) which is a neural network subjected to learning.Type: ApplicationFiled: August 16, 2010Publication date: July 19, 2012Applicant: SHARP KABUSHIKI KAISHAInventors: Takahiro Sasaki, Satoru Kishida, Kentaro Kinoshita
-
Patent number: 8208697Abstract: A method for determining the presence or absence of malignant features in medical images, wherein a plurality of base comparison or training images of various types of lesions taken of actual patient is examined by one or more image reading experts to create a first database array. Low-level features of each of the lesions in the same plurality of base comparisons or training images are determined using one or more image processing algorithms to obtain a second database array set. The first and second database array set are combined to create a training database array set which is input to a learning system that discovers/learns a classifier that maps from a subset of the low-level features to the expert's evaluation in the first database array set. The classifier is used to determine the presence of a particular mid-level feature in an image of lesion in a patient based solely on the image.Type: GrantFiled: December 13, 2005Date of Patent: June 26, 2012Assignee: Koninklijke Philips Electronics N.V.Inventors: James David Schaffer, Walid Ali, Larry J. Eshelman, Claude Cohen-Bacrie, Jean-Michel Lagrange, Claire Levrier, Nicholas Villain, Robert R. Entrekin
-
Patent number: 8175375Abstract: A method of compression of videotelephony images characterized by: creating (10) a learning base containing images; centering the learning base about zero; determining component images by principal component analysis (12); and keeping a number of significant principal components (14).Type: GrantFiled: December 29, 2002Date of Patent: May 8, 2012Assignee: Texas Instruments IncorporatedInventors: Christophe Bonnery, Jean-Yves Desbree, Chrisophe Flouzat, Daniel Le Guennec, David Mercier, Mickaël Remingol, Renaud Seguier, David Thomas, Gilles Vaucher
-
Patent number: 8170343Abstract: A method and a system for searching images with figures and a recording medium storing metadata of image are provided. The searching method is divided into an image analysis stage and an image search stage. In the image analysis stage, figures between images are compared with each other and assigned with an identity respectively. A representative image of each identity is then evaluated from the image collection. In the image search stage, the representative images are displayed for user to select some of them as a searching criterion, so as to search and display the images matching the searching criterion in the image collection. Accordingly, the images required by user can be found through intelligent analysis of figures, intuitive definition of searching criterion, and simple comparison of identities so that both time and effort of organization for searching images with figures can be substantially saved.Type: GrantFiled: September 2, 2007Date of Patent: May 1, 2012Assignee: Corel CorporationInventor: Yi-Tsung Chien
-
Patent number: 8160354Abstract: An image-based pattern recognizer and a method and apparatus for making such a pattern recognizer are disclosed. By employing positional coding, the meaning of any feature present in an image can be defined implicitly in space. The pattern recognizer can be a neural network including a plurality of stages of observers. The observers are configured to cooperate to identify the presence of features in the input image and to recognize a pattern in the input image based on the features. Each of the observers includes a plurality of neurons. The input image includes a plurality of units, and each of the observers is configured to generate a separate output set that includes zero or more coordinates of such units.Type: GrantFiled: December 26, 2008Date of Patent: April 17, 2012Assignee: Five Apes, Inc.Inventor: Williams J. F. Paquier
-
Patent number: 8131065Abstract: Machine-readable media, methods, apparatus and system for obtaining and processing image features are described. In some embodiments, groups of training features derived from regions of training images may be trained to obtain a plurality of classifiers, each classifier corresponding to each group of training features. The plurality of classifiers may be used to classify groups of validation features derived from regions of validation images to obtain a plurality of weights, wherein each weight corresponds to each region of the validation images and indicates how important the each region of the validation images is. Then, a weight may be discarded from the plurality of weights based upon a certain criterion.Type: GrantFiled: December 20, 2007Date of Patent: March 6, 2012Assignee: Intel CorporationInventors: Jianguo Li, Tao Wang, Yimin Zhang
-
Patent number: 8130827Abstract: A method and apparatus are disclosed for interpolating an object reference pixel in an annular image. In one embodiment, reference pixels selected based on a distorted shape of the annular image are arranged in a direction of distortion of the annular image and an object reference pixel in the annular image is interpolated based on the selected reference pixels.Type: GrantFiled: August 11, 2005Date of Patent: March 6, 2012Assignees: Samsung Electronics Co., Ltd., Industry Academic Cooperation Foundation Kyunghee UniversityInventors: Gwang-Hoon Park, Sung-Keun Kim
-
Publication number: 20120045119Abstract: A plurality of features determined from at least a portion of an image containing information about an object are processed with an inclusive neural network, and with a plurality of exclusive neural networks, so as to provide a plurality of inclusive probability values representing probabilities that the portion of the image corresponds to at least one of at least two different classes of objects, and for each exclusive neural network, so as to provide first and second exclusive probability values representing probabilities that the portion of the image respectively corresponds. or not. to at least one class of objects. The plurality of inclusive probability values, and the first and second exclusive probability values from each of the exclusive neural networks, provide for identifying whether the portion of the image corresponds, or not, to any of the at least two different classes of objects.Type: ApplicationFiled: November 1, 2011Publication date: February 23, 2012Applicant: AUTOMOTIVE SYSTEMS LABORATORY, INC.Inventor: Gregory G. SCHAMP
-
Patent number: 8103091Abstract: It is to learn an object identification parameter while suppressing an influence of the background area. The object identification parameter learning system includes: a feature extracting device for obtaining a feature of an object from the image; a background specifying device for specifying a background area of the image; a background replacing device which replaces feature components corresponding to the background area of the feature with other values; and an identification parameter update device for updating the identification parameter based on the feature components replaced by the background replacing device. The identification parameter can be learnt by generating a plurality of pieces of feature data of the object with different backgrounds from a single object image through replacing the background area of the feature of the object.Type: GrantFiled: August 27, 2007Date of Patent: January 24, 2012Assignee: NEC CorporationInventor: Toshinori Hosoi
-
Patent number: 8041441Abstract: A production management system has processing devices A, B, C, D, E, and P. A kind of product ? is processed in the order of the processing devices A, P, B, P, and C, and a kind of product ? is processed in the order of the processing devices D, P, E, and P. To determine whether the processing device P is to be used to produce the product ? or the product ?, an input ratio of each kind of product is multiplied by the number of times of passing the processing device P for each kind of product, thereby calculating a core of each kind of product. Based on the calculated score, whether the processing device P is to be used to produce the product ? or the product ? is determined. Accordingly, the work-in-process balance of key processes between different kinds of products can be equalized.Type: GrantFiled: January 25, 2008Date of Patent: October 18, 2011Assignee: Elpida Memory, Inc.Inventors: Hiroaki Izumi, Katsuhiko Takahashi, Katsumi Morikawa
-
Patent number: 8031914Abstract: Face-based image clustering systems and methods are described. In one aspect, face regions are detected in images. At least one respective parameter value is extracted from each of the face regions. Ones of the face regions associated with parameter values satisfying a cluster seed predicate are classified as cluster seed face regions. The cluster seed face regions are clustered into one or more clusters. A respective face model is built for each of the clusters. The face models are stored. In another aspect, face regions are detected in images. At least one respective parameter value is extracted from each of the face regions. The face regions are ranked based on the extracted parameter values. The face regions are clustered in rank order into one or more clusters. Representations of ones of the clusters are rendered on a display.Type: GrantFiled: October 11, 2006Date of Patent: October 4, 2011Assignee: Hewlett-Packard Development Company, L.P.Inventor: Tong Zhang
-
Patent number: 7995837Abstract: The disclosed terrain model is a generative, probabilistic approach to modeling terrain that exploits the 3D spatial structure inherent in outdoor domains and an array of noisy but abundant sensor data to simultaneously estimate ground height, vegetation height and classify obstacles and other areas of interest, even in dense non-penetrable vegetation. Joint inference of ground height, class height and class identity over the whole model results in more accurate estimation of each quantity. Vertical spatial constraints are imposed on voxels within a column via a hidden semi-Markov model. Horizontal spatial constraints are enforced on neighboring columns of voxels via two interacting Markov random fields and a latent variable. Because of the rules governing abstracts, this abstract should not be used to construe the claims.Type: GrantFiled: October 5, 2009Date of Patent: August 9, 2011Assignee: Carnegie Mellon UniversityInventors: Carl K. Wellington, Aaron C. Courville, Anthony J. Stentz
-
Patent number: RE43896Abstract: The present invention provides an image and video indexing scheme for content analysis. According to the invention, a database of images or videos is compressed. By examining patterns in the compression scheme of each image or video, the present invention identifies the content of the data. In one embodiment, an unsupervised learning method is employed where each image or video is sub-divided into smaller blocks (8 pixels×8 pixels, for instance) and each of the smaller blocks is examined for its compression pattern. Then, the patterns associated with each of the smaller blocks is recorded for each of the images in the database and content is retrieved from the database by associating certain patterns or groups of patterns with certain content.Type: GrantFiled: May 16, 2008Date of Patent: January 1, 2013Assignee: California Institute of TechnologyInventors: Patricia A. Keaton, Rodney M. Goodman