Local Neighborhood Operations (e.g., 3x3 Kernel, Window, Or Matrix Operator) Patents (Class 382/205)
  • Patent number: 10521697
    Abstract: A local connectivity feature transform (LCFT) is applied to binary document images containing text characters, to generate transformed document images which are then input into a bi-directional Long Short Term Memory (LSTM) neural network to perform character/word recognition. The LCFT transformed image is a gray scale image where the pixel values encode local pixel connectivity information of corresponding pixels in the original binary image. The transform is one that provides a unique transform score for every possible shape represented as a 3×3 block. In one example, the transform is computed using a 3×3 weight matrix that combines bit coding with a zigzag pattern to assign weights to each element of the 3×3 block, and by summing up the weights for the non-zero elements of the 3×3 block shape.
    Type: Grant
    Filed: September 29, 2017
    Date of Patent: December 31, 2019
    Assignee: KONICA MINOLTA LABORATORY U.S.A., INC.
    Inventors: Shubham Agarwal, Maral Mesmakhosroshahi, Yongmian Zhang
  • Patent number: 10504206
    Abstract: A method and system for enhancing spatial resolution of an image of an object includes using an image detector with a sensor array of pixels configured for extended dynamic range detection. Multiple image sample are collected to assemble a stack of image samples and input into an image processor for processing using an image reconstruction engine to generate a single output image from the stack. In a preferred embodiment, the image reconstruction engine executes the Pixon method for image reconstruction.
    Type: Grant
    Filed: July 6, 2017
    Date of Patent: December 10, 2019
    Assignee: Micro Biz Ventures LLC
    Inventor: Richard C. Puetter
  • Patent number: 10482336
    Abstract: Systems and methods are provided for performing automated face recognition and comparison. An input image of a face may be received and cropped, and the image may be processed through a deep neural network (DNN) to produce a k-dimensional feature vector. The k-dimensional feature vector may be converted to a k-dimensional binary vector by transforming each value in the vector to either 1 or 0. To search for nearest matches of the image in a database of gallery images of faces, the system may compare sub-strings of the binary vector to hash tables created from sub-strings of the gallery images, enabling sub-linear searching that allows locating the closest matches from among the entire gallery without requiring an exhaustive linear search of the entire gallery.
    Type: Grant
    Filed: October 4, 2017
    Date of Patent: November 19, 2019
    Assignee: NOBLIS, INC.
    Inventors: Mark J. Burge, Jordan Cheney
  • Patent number: 10412398
    Abstract: Provided is an image processing device including a first rearrangement unit configured to rearrange first divided image data, which is image data corresponding to respective first divided regions obtained by dividing images to be processed represented by processing target image data in the horizontal direction and in the vertical direction, for each of second divided regions, obtained by dividing the images to be processed composed of a plurality of the first divided regions, in an order corresponding to the respective second divided regions, a compression processing unit configured to compress respective pieces of second divided image data, which are image data corresponding to the respective second divided regions, by performing a transform in a predetermined scheme, quantization, and variable length encoding on the data, and a second rearrangement unit configured to rearrange the compressed second divided image data in an order corresponding to all of the images to be processed.
    Type: Grant
    Filed: February 24, 2015
    Date of Patent: September 10, 2019
    Assignee: SONY CORPORATION
    Inventor: Kaoru Urata
  • Patent number: 10346023
    Abstract: One or more embodiments increase the visual appeal and organization of presenting digital content items. In particular, one or more embodiments improve the visual appeal of a presentation of digital content items by emphasizing one or more digital content items within the presentation of digital content items. For instance, one or more digital content items may be emphasized within a graphical user interface including a gallery of digital content items. The emphasized digital content items can be selected based on a score reflecting a particular characteristic of the digital content item and/or a metric of user interest in the digital content item. Digital content items with relatively high scores as compared to other digital content items may be emphasized when presented in conjunction with the other digital content items.
    Type: Grant
    Filed: April 8, 2014
    Date of Patent: July 9, 2019
    Assignee: DROPBOX, INC.
    Inventors: George Milton Underwood, IV, Ramesh Balakrishnan, Boris Babenko, Christopher Lee, Brian Smith
  • Patent number: 10198671
    Abstract: A dense captioning system and method is provided for processing an image to produce a feature map of the image, analyzing the feature map to generate proposed bounding boxes for a plurality of visual concepts within the image, analyzing the feature map to determine a plurality of region features of the image, and analyzing the feature map to determine a context feature for the image. For each region feature of the plurality of region features of the image, the dense captioning system further provides for analyzing the region feature to determine a detection score for the region feature, calculating a caption for a bounding box for a visual concept in the image using the region feature and the context feature, and localizing the visual concept by adjusting the bounding box around the visual concept based on the caption to generate an adjusted bounding box for the visual concept.
    Type: Grant
    Filed: November 10, 2016
    Date of Patent: February 5, 2019
    Assignee: Snap Inc.
    Inventors: Linjie Yang, Kevin Dechau Tang, Jianchao Yang, Jia Li
  • Patent number: 10084779
    Abstract: Embodiments are directed to using a hash signature of a rendered DOM object of a website to find similar content and behavior on other websites. Embodiments break a DOM into a large number of data portions (i.e., “shingles”), apply a hashing algorithm to the shingles, select a predetermined number of hashes from the hashed shingles according to a selection criteria to create a hash signature, and compare the hash signature to that of a reference page to determine similarity of website DOM object content. Embodiments can be used to identify phishing websites, defaced websites, spam websites, significant changes in the content of a webpage, copyright infringement, and any other suitable purposes related to the similarity between website DOM object content.
    Type: Grant
    Filed: June 16, 2017
    Date of Patent: September 25, 2018
    Assignee: RiskIQ, Inc.
    Inventors: Adam Hunt, David Pon, Chris Kiernan, Ben Adams, Jonas Edgeworth, Elias Manousos, Joseph Linn
  • Patent number: 10075290
    Abstract: A system for performing an operation on data using obfuscated representations of the data is disclosed. Obtaining means are configured to obtain a first obfuscated representation of a first data value and obtain a second obfuscated representation of a second data value. A determining means 102 is configured to determine an obfuscated representation of a third data value, by performing the corresponding operations on the obfuscated representation of the first data value and the obfuscated representation of the second data value. Obfuscating means 101 may be configured to generate the first obfuscated representation based on the first data value and generate the second obfuscated representation based on the second data value. De-obfuscating means 103 may be configured to de-obfuscate the obfuscated representation of the third data value in order to obtain the third data value using a system of equations.
    Type: Grant
    Filed: December 17, 2014
    Date of Patent: September 11, 2018
    Assignee: KONINKLIJKE PHILIPS N.V.
    Inventors: Paulus Mathias Hubertus Mechtildis Antonius Gorissen, Ludovicus Marinus Gerardus Maria Tolhuizen
  • Patent number: 10025755
    Abstract: A method and apparatus for processing data are provided. The processor includes an input buffer, a data extractor, a multiplier, and an adder. The input buffer receives data and stores the data. The data extractor extracts kernel data corresponding to a kernel in the data from the input buffer. The multiplier multiplies the extracted kernel data by a convolution coefficient. The adder calculates a sum of multiplication results from the multiplier.
    Type: Grant
    Filed: January 7, 2016
    Date of Patent: July 17, 2018
    Assignee: SAMSUNG ELECTRONICS CO., LTD.
    Inventors: Moradi Saber, Jun Haeng Lee, Eric Hyunsurk Ryu, Keun Joo Park
  • Patent number: 10020033
    Abstract: Systems and methods are provided for managing access to registers. In one embodiment, a system may include a processor and a plurality of registers. The processor and the plurality of registers may be integrated into a single device, or may be in separate devices. The plurality of registers may include a first set of registers that are directly accessible by the processor, and a second set of registers that are not directly accessible by the processor. The second set of registers may, however, be accessed indirectly by the processor via the first set of registers. In one embodiment, the first set of registers may include a register for selecting a register bank from the second set of registers, and a register for selecting a particular address within the register bank, to allow indirect access by the processor to the registers of the second set.
    Type: Grant
    Filed: August 14, 2017
    Date of Patent: July 10, 2018
    Assignee: Micron Technology, Inc.
    Inventors: Harold B Noyes, Mark Jurenka, Gavin Huggins
  • Patent number: 9940731
    Abstract: Asymmetries are detected in one or more images by partitioning each image to create a set of patches. Salient patches are identified, and an independent displacement for each patch is identified. The techniques used to identify the salient patches and the displacement for each patch are combined in a function to generate a score for each patch. The scores can be used to identify possible asymmetries.
    Type: Grant
    Filed: September 4, 2015
    Date of Patent: April 10, 2018
    Assignee: International Business Machines Corporation
    Inventors: Sharon Alpert, Miri Erihov, Pavel Kisilev
  • Patent number: 9734876
    Abstract: Systems and methods are provided for managing access to registers. In one embodiment, a system may include a processor and a plurality of registers. The processor and the plurality of registers may be integrated into a single device, or may be in separate devices. The plurality of registers may include a first set of registers that are directly accessible by the processor, and a second set of registers that are not directly accessible by the processor. The second set of registers may, however, be accessed indirectly by the processor via the first set of registers. In one embodiment, the first set of registers may include a register for selecting a register bank from the second set of registers, and a register for selecting a particular address within the register bank, to allow indirect access by the processor to the registers of the second set.
    Type: Grant
    Filed: January 19, 2015
    Date of Patent: August 15, 2017
    Assignee: Micron Technology, Inc.
    Inventors: Harold B Noyes, Mark Jurenka, Gavin Huggins
  • Patent number: 9729755
    Abstract: As set forth herein, methods and systems for automated image defect detection. An input image is subjected to skew detection and contaminant detection. Upon detection of skew of and/or contaminants in the image data, correction data, including indicia of the contaminants/skew, correction angles and directions are calculated and displayed on an electronic representation of the document. Automatic correction may then be performed to correct the detected defects, allow the defects to remain, or re-scan the original document to generate new image data without the detected defects.
    Type: Grant
    Filed: August 31, 2011
    Date of Patent: August 8, 2017
    Assignee: Xerox Corporation
    Inventors: Thomas William Stuart, Asad Khalid Raja, Richard Henry Bay, Samuel Luke Mills
  • Patent number: 9686283
    Abstract: Embodiments are directed to using a hash signature of a rendered DOM object of a website to find similar content and behavior on other websites. Embodiments break a DOM into a large number of data portions (i.e., “shingles”), apply a hashing algorithm to the shingles, select a predetermined number of hashes from the hashed shingles according to a selection criteria to create a hash signature, and compare the hash signature to that of a reference page to determine similarity of website DOM object content. Embodiments can be used to identify phishing websites, defaced websites, spam websites, significant changes in the content of a webpage, copyright infringement, and any other suitable purposes related to the similarity between website DOM object content.
    Type: Grant
    Filed: May 20, 2016
    Date of Patent: June 20, 2017
    Assignee: RiskIQ, Inc.
    Inventors: Adam Hunt, David Pon, Chris Kiernan, Ben Adams, Jonas Edgeworth, Elias Manousos, Joseph Linn
  • Patent number: 9639948
    Abstract: Methods, apparatuses, and computer readable storage media are provided for determining a depth measurement of a scene using an optical blur difference between two images of the scene. Each image is captured using an image capture device with different image capture device parameters. A corresponding image patch is identified from each of the captured images, motion blur being present in each of the image patches. A kernel of the motion blur in each of the image patches is determined. The kernel of the motion blur in at least one images patch is used to generate a difference convolution kernel. A selected first image patch is convolved with the generated difference convolution kernel to generate a modified image patch. A depth measurement of the scene is determined from an optical blur difference between the modified image patch and the remaining image patch.
    Type: Grant
    Filed: December 22, 2014
    Date of Patent: May 2, 2017
    Assignee: Canon Kabushiki Kaisha
    Inventors: David Peter Morgan-Mar, Matthew Raphael Arnison
  • Patent number: 9530069
    Abstract: Various embodiments of the present invention relate to a method, system and computer program product for detecting and recognizing text in the images captured by cameras and scanners. First, a series of image-processing techniques is applied to detect text regions in the image. Subsequently, the detected text regions pass through different processing stages that reduce blurring and the negative effects of variable lighting. This results in the creation of multiple images that are versions of the same text region. Some of these multiple versions are sent to a character-recognition system. The resulting texts from each of the versions of the image sent to the character-recognition system are then combined to a single result, wherein the single result is detected text.
    Type: Grant
    Filed: February 3, 2015
    Date of Patent: December 27, 2016
    Assignee: A9.com, Inc.
    Inventors: Raghavan Manmatha, Mark A. Ruzon
  • Patent number: 9524029
    Abstract: The present invention is to appropriately determine a gesture operation detected to perform data processing according thereto. In the present invention, a CPU judges a processing status at the time of the detection of a gesture operation performed on a touch panel, and after determining its gesture operation type according to the processing status, performs data processing according to the operation type. In this case, the CPU determines and evaluates a gesture operation type for each processing status, and updates values (evaluation accumulation values) of a flick priority pointer and a tap priority pointer. Then, based on these values of the flick priority pointer and the tap priority pointer, the CPU determines one of the gesture operation types (flick operation/tap operation).
    Type: Grant
    Filed: July 16, 2013
    Date of Patent: December 20, 2016
    Assignee: CASIO COMPUTER CO., LTD
    Inventor: Hiroki Kanno
  • Patent number: 9426382
    Abstract: A first image captured using a first exposure and a second image captured using a second exposure higher than the first exposure are obtained. A third image used as a reference for combining processing is created from the first image. An image, a dynamic range of which is extended, is created by weighted-combining the first and second images using adding coefficients, which are associated with pixel values of the third image in advance.
    Type: Grant
    Filed: January 26, 2012
    Date of Patent: August 23, 2016
    Assignee: Canon Kabushiki Kaisha
    Inventor: Masaaki Matsuoka
  • Patent number: 9386037
    Abstract: Embodiments are directed to using a hash signature of a rendered DOM object of a website to find similar content and behavior on other websites. Embodiments break a DOM into a large number of data portions (i.e., “shingles”), apply a hashing algorithm to the shingles, select a predetermined number of hashes from the hashed shingles according to a selection criteria to create a hash signature, and compare the hash signature to that of a reference page to determine similarity of website DOM object content. Embodiments can be used to identify phishing websites, defaced websites, spam websites, significant changes in the content of a webpage, copyright infringement, and any other suitable purposes related to the similarity between website DOM object content.
    Type: Grant
    Filed: November 11, 2015
    Date of Patent: July 5, 2016
    Assignee: RiskIQ Inc.
    Inventors: Adam Hunt, David Pon, Chris Kiernan, Ben Adams, Jonas Edgeworth, Elias Manousos, Joseph Linn
  • Patent number: 9377934
    Abstract: An image display unit includes an image display module changing for display images each having image and header information, a touch panel input module receiving inputs, an image reading module reading the image and header information of the images, a group definition determination module determining, based on the header information of a first image of the images that is displayed next in response to an operation performed on the touch panel input module, whether or not a group definition in the header information of the first image is the same as a first predetermined group definition, and an image changing module changing the images to the first image for display in a number of steps predetermined for the first predetermined group definition.
    Type: Grant
    Filed: February 26, 2013
    Date of Patent: June 28, 2016
    Assignee: CASIO COMPUTER CO., LTD.
    Inventor: Hiroki Kanno
  • Patent number: 9344697
    Abstract: A method for transitioning between an image of a first video sequence and an image of a second video sequence, wherein an image of the video sequences corresponds to one sub-region of a respective image of the other video sequence, receiving one image from the first video sequence; providing, from the one image, one pre-transition image; for each one pre-transition image: determining a confidence region within the one pre-transition image based on a set of confidence values, each confidence value representing, for a respective block of the one pre-transition image, the level of reliability of determining one or more blocks of a corresponding image in the second video sequence; computing a transition image corresponding to the second video sequence, the transition image being a function of the determined confidence region of the pre-transition image; and displaying the transition image for a duration upon the reception of the second video sequence.
    Type: Grant
    Filed: February 5, 2013
    Date of Patent: May 17, 2016
    Assignee: Canon Kabushiki Kaisha
    Inventors: Naël Ouedraogo, Franck Denoual, Fabrice Le Leannec
  • Patent number: 9240037
    Abstract: An image processing apparatus includes: a reduced image generation unit configured to obtain a reduced image; a gain data generation unit configured to obtain gain data; and a determination unit configured to determine a control gain required to amplify the input image based on the generated gain data, and letting first gain data be the gain data for an image having a relatively high frequency, and second gain data be the gain data having a relatively low frequency, the determination unit determines the control gain based on a first addition result and a second addition result.
    Type: Grant
    Filed: February 10, 2014
    Date of Patent: January 19, 2016
    Assignee: Canon Kabushiki Kaisha
    Inventor: Shota Yamaguchi
  • Patent number: 9189827
    Abstract: System for conveying a stream of information. The system includes an encoding device employing a spatio-temporal coding scheme that omits light, including codes embedded therein that are invisible to a user. A receiver that might be a cell phone camera receives light from the encoding device and computer apparatus is programmed with software to decode the received light to generate the stream of information. The encoding device is preferably a video display.
    Type: Grant
    Filed: October 1, 2012
    Date of Patent: November 17, 2015
    Assignee: Massachusetts Institute of Technology
    Inventors: Grace Rusi Woo, Andrew B Lippman
  • Patent number: 9047655
    Abstract: A system and method of symbol matching may include a processor configured to determine which pixels of a first symbol are tangent pixels; for each of the determined tangent pixels, determine whether a second symbol includes a pixel corresponding to the tangent pixel that includes at least one same tangent constraint as that of the tangent pixel; accept the first and second symbols as a match based on shared tangent constraints conditional upon a determination that the second symbol includes for each of at least a subset of the tangent pixels of the first symbol a corresponding pixel that includes the at least one same tangent constraint as that of the tangent pixel; and generate a document including a single symbol that is mapped to the first and second symbols.
    Type: Grant
    Filed: February 10, 2014
    Date of Patent: June 2, 2015
    Assignee: CVision Technologies, Inc.
    Inventors: Ari David Gross, Raphael Meyers, Navdeep Tinna
  • Publication number: 20150131915
    Abstract: In techniques for adaptive denoising with internal and external patches, example image patches taken from example images are grouped into partitions of similar patches, and a partition center patch is determined for each of the partitions. An image denoising technique is applied to image patches of a noisy image to generate modified image patches, and a closest partition center patch to each of the modified image patches is determined. The image patches of the noisy image are then classified as either a common patch or a complex patch of the noisy image, where an image patch is classified based on a distance between the corresponding modified image patch and the closest partition center patch. A denoising operator can be applied to an image patch based on the classification, such as applying respective denoising operators to denoise the image patches that are classified as the common patches of the noisy image.
    Type: Application
    Filed: November 14, 2013
    Publication date: May 14, 2015
    Applicant: Adobe Systems Incorporated
    Inventors: Zhe Lin, Jianchao Yang, Hailin Jin, Xin Lu
  • Patent number: 9031291
    Abstract: A ridge flow based fingerprint image quality determination can be achieved independent of image resolution, can be processed in real-time and includes segmentation, such as fingertip segmentation, therefore providing image quality assessment for individual fingertips within a four finger flat, dual thumb, or whole hand image. A fingerprint quality module receives from one or more scan devices ridge-flow—containing imagery which can then be assessed for one or more of quality, handedness, historical information analysis and the assignment of bounding boxes.
    Type: Grant
    Filed: March 22, 2013
    Date of Patent: May 12, 2015
    Assignee: Aware, Inc.
    Inventors: James M. Maurer, L. Scott Hills
  • Patent number: 9021347
    Abstract: An information processing method comprises the steps of: inputting an input information of a multi-dimensional array; calculating an accumulated information value corresponding to a position of each element of the input information; and holding the accumulated information value in a buffer having a size of predetermined bits, wherein in the holding step, when an accumulated information value calculated in the calculating step overflows with respect to the size, a part not more than the predetermined bits of the calculated accumulated information value is held as the accumulated information value.
    Type: Grant
    Filed: May 27, 2008
    Date of Patent: April 28, 2015
    Assignee: Canon Kabushiki Kaisha
    Inventors: Yoshinori Ito, Kan Torii, Takahisa Yamamoto, Masami Kato
  • Patent number: 9014483
    Abstract: Input information of a multidimensional array is divided into a plurality of divided areas, accumulated information is generated by calculating accumulated values at respective element positions of the input information from a corresponding reference location for each of the plurality of divided areas, and the generated accumulated information is held in a memory for each divided area. Calculation using the accumulated information is executed for a predetermined processing range. The input information is divided into the plurality of divided areas so that two neighboring divided areas have an overlapping area, and the overlapping area has a size at least in which the whole processing range fits.
    Type: Grant
    Filed: August 31, 2010
    Date of Patent: April 21, 2015
    Assignee: Canon Kabushiki Kaisha
    Inventors: Yoshinori Ito, Takahisa Yamamoto, Masami Kato
  • Publication number: 20150103199
    Abstract: Compact descriptors of digital images are produced by detecting interest points representative of the digital images and selecting out of the interest points key points for producing e.g. local and global compact descriptors of the images. The digital images are decomposed into blocks by computing an energy (variance) for each said block and then subjecting the blocks to culling by rejecting those blocks having an energy failing to pass an energy threshold. The interest points are detected only in the blocks resulting from culling, and the key points for producing the compact descriptors are selected out of the interest points thus detected, possibly by using different selection thresholds for local and global compact descriptors, respectively. The number of key points for producing the compact descriptors may be varied e.g. by adaptively varying the number of the interest points detected in the blocks resulting from culling.
    Type: Application
    Filed: October 16, 2014
    Publication date: April 16, 2015
    Inventor: Danilo Pietro Pau
  • Patent number: 9002055
    Abstract: A method (100) and system (300) is described for processing video data comprising a plurality of images. The method (100) comprising obtaining (104, 106), for each of the plurality of images, a segmentation in a plurality of regions and a set of keypoints, and tracking (108) at least one region between a first image and a subsequent image resulting in a matched region in the subsequent image taking into account a matching between keypoints in the first image and the subsequent image. The latter results in accurate tracking of regions. Furthermore the method may optionally also perform label propagation taking into account keypoint tracking.
    Type: Grant
    Filed: October 13, 2008
    Date of Patent: April 7, 2015
    Assignees: Toyota Motor Europe NV, Cambridge Enterprise Limited
    Inventors: Ryuji Funayama, Hiromichi Yanagihara, Julien Fauqueur, Gabriel Brostow, Roberto Cipolla
  • Patent number: 9002118
    Abstract: An image feature extraction device according to an embodiment includes a gradient image calculator generates intensity gradient data with respect to two different directions based on intensity data of image data; and a gradient count unit calculates a covariance matrix for each partial area obtained by dividing the image data based on the intensity gradient data. The image feature extraction device according to the embodiment further includes a feature data output unit calculates two parameters related to a major axis and a minor axis of an ellipse expressed by the covariance matrix, quantizes a range of the logarithms of the parameters for each of the partial area using a predetermined division number, and outputs a feature vector which contains a value only at a dimension corresponding to the quantized range different from the other dimensions.
    Type: Grant
    Filed: September 11, 2013
    Date of Patent: April 7, 2015
    Assignee: Kabushiki Kaisha Toshiba
    Inventor: Nobuhiro Nonogaki
  • Patent number: 8995758
    Abstract: According to an embodiment, a method for filtering descriptors for visual object recognition is provided. The method includes identifying false positive descriptors having a local match confidence that exceeds a predetermined threshold and a global image match confidence that is less than a second threshold. The method also includes training at least one classifier to discriminate between the false positive descriptors and other descriptors. The method further includes filtering feature point matches using the at least one classifier. According to another embodiment, the filtering step may further include removing one or more feature point matches from a result set. According to a further embodiment, a system for filtering feature point matches for visual object recognition is provided. The system includes a hard false positive identifier, a classifier trainer and a hard false positive filter.
    Type: Grant
    Filed: June 22, 2009
    Date of Patent: March 31, 2015
    Assignee: Google Inc.
    Inventors: Alessandro Bissacco, Ulrich Buddemeier, Hartmut Neven
  • Patent number: 8988732
    Abstract: In the image processing apparatus, image data is divided into large blocks of a prescribed size and the large blocks are subdivided into small blocks by the dividing unit. The number of isolated points in each large block is then calculated by the large block isolated point calculation unit, and the number of isolated points in each small block is then calculated by the small block isolated point calculation units. It is then determined by the halftone-dot region determination unit whether or not the large block is a halftone-dot region. This determination considers both the number of isolated points in the large block and the number of isolated points in each small block.
    Type: Grant
    Filed: September 16, 2003
    Date of Patent: March 24, 2015
    Assignee: Konica Minolta Business Technologies, Inc.
    Inventors: Tomohiro Yamaguchi, Yoshihiko Hirota
  • Publication number: 20150055837
    Abstract: Discerning small, weak features (“specks”) from imagery can be critical in early-stage cancer detection and other applications. In one aspect, the presence of a speck is judged from a large set of “votes” about whether a point in the imagery has a value above or below its neighbors. Hundreds or thousands or more votes can be gathered from one or more images, to thereby—in the aggregate—tend to confirm or refute the presence of a speck at a particular location. Votes can be based on simple pixel differences, e.g., between a pixel at the center of a 7×7 pixel image excerpt, and each of 48 other pixels in the excerpt. More sophisticated methods can employ analyses of triads, quads, and rings, and kernel-based Markov Random Field frameworks, to roll-up to a final conclusion. A great number of other features and arrangements are also detailed. The technology is particularly illustrated in the context of detecting exoplanets from astronomical imagery of remote star systems.
    Type: Application
    Filed: August 22, 2014
    Publication date: February 26, 2015
    Inventor: Geoffrey B. Rhoads
  • Publication number: 20150049943
    Abstract: In general, techniques are described for performing a vocabulary-based visual search using multi-resolution feature descriptors. A device may comprise one or more processors configured to perform the techniques. The one or more processors may to apply a partitioning algorithm to a first subset of target feature descriptors to determine a first classifying data structure to be used when performing a visual search with respect to a query feature descriptor. The one or more processors may then apply the partitioning algorithm to a second subset of the target feature descriptors to determine a second classifying data structure to be used when performing the visual search with respect to the same query feature descriptor.
    Type: Application
    Filed: September 30, 2013
    Publication date: February 19, 2015
    Applicant: QUALCOMM Incorporated
    Inventor: Onur C. Hamsici
  • Publication number: 20150049942
    Abstract: In general, techniques are described for performing a vocabulary-based visual search using multi-resolution feature descriptors. A device may comprise one or more processors configured to perform the techniques. The processors may generate a hierarchically arranged data structure to be used when classifying objects included within a query image based on multi-resolution query feature descriptor extracted from the query image at a first scale space resolution and a second scale space resolution. The hierarchically arranged data structure may represent a first query feature descriptor of the multi-resolution feature descriptor extracted at the first scale space resolution and a second corresponding query feature descriptor of the multi-resolution feature descriptor extracted at the second scale space resolution hierarchically arranged according to the first scale space resolution and the second scale space resolution. The processors may then perform a visual search based on the generated data structure.
    Type: Application
    Filed: September 30, 2013
    Publication date: February 19, 2015
    Applicant: QUALCOMM Incorporated
    Inventor: Onur C. Hamsici
  • Patent number: 8958661
    Abstract: Methods and apparatus to generate templates from web images for searching an image database are described. In one embodiment, one or more retrieved images (e.g., from the Web) may be used to generate one or more templates. The templates may be used to search an image database based on features commonly shared between sub-images of the retrieved images. Other embodiments are also described.
    Type: Grant
    Filed: March 30, 2007
    Date of Patent: February 17, 2015
    Assignee: Intel Corporation
    Inventors: Navneet Panda, Yi Wu, Jean-Yves Bouguet, Ara Nefian
  • Patent number: 8958662
    Abstract: Methods and systems for automating insertion of content into media-based projects are disclosed. In one embodiment, images are uploaded from a user device to a server. The server performs image analysis on the uploaded images and obtains automatic image information. After presentation of the initial project, the user manually interacts with the uploaded images to provide user inputs to the project. The system observes the user interactions to generate manual image information based on the user interactions with the project. Based on the automatic image information and the manual image information, the present invention automatically updates a printable photobook or other media-based project after automatically inserting relevant content into the project. This produces a faster and more seamless user experience, and increases the number of customers purchasing the ultimate product by minimizing friction points to purchase.
    Type: Grant
    Filed: August 20, 2014
    Date of Patent: February 17, 2015
    Assignee: Interactive Memories, Inc.
    Inventors: Aryk Erwin Grosz, Anthony Jan Fung Lai, Dan Schultz, Denys Barushev, Anton Vladimirovich Astashov, Wiley Wang
  • Patent number: 8953903
    Abstract: A method of multi-frame image noise reduction suitable for an image-capturing device includes following steps: obtaining a current frame and multiple reference frames; defining a mask and a target point in the mask; judging whether the target point pixel of the current frame is on an edge according to an edge map of the frame; when the pixel is on the edge, using the pixels in the reference frames on the edge to calculate a replacement result; when the target point pixel is not on the edge, using the pixels in the reference frames surrounding the target point to calculate a replacement result; after that, generating a pixel corresponding to the position of the target point in an output image according to the replacement result; further, moving the mask and going back to judge whether the pixel of the target point of the current frame is on the edge.
    Type: Grant
    Filed: August 9, 2011
    Date of Patent: February 10, 2015
    Assignee: Altek Corporation
    Inventors: Shih-Yuan Peng, Chung-Ta Wu
  • Publication number: 20150010240
    Abstract: An image processing apparatus for searching for corresponding points between a first image and a second image is provided. The image processing apparatus includes a window setting portion, a phase information extraction portion that extracts phase information by subjecting a representation included in each set window to frequency decomposition, a corresponding points determination portion that determines the corresponding points based on comparison of the phase information extracted from each window, and a window size control portion that causes the window setting portion to set a window different in size in accordance with a frequency used for calculation of the frequency decomposition. The window size control portion causes the window setting portion to set a window having a first size in response to a first frequency and to set a window having a second size smaller than the first size in response to a second frequency higher than the first frequency.
    Type: Application
    Filed: December 26, 2012
    Publication date: January 8, 2015
    Applicant: KONICA MINOLTA, INC.
    Inventor: Hironori Sumitomo
  • Publication number: 20140376819
    Abstract: One or more aspects of the subject disclosure are directed towards identifying objects within an image via image searching/matching. In one aspect, an image is processed into bounding boxes, with the bounding boxes further processed to each surround a possible object. A sub-image of pixels corresponding to the bounding box is featurized for matching with tagged database images. The information (tags) associated with any matched images is processed to identify/categorize the sub-image and thus the object corresponding thereto.
    Type: Application
    Filed: June 21, 2013
    Publication date: December 25, 2014
    Inventors: Ce Liu, Yair Weiss, Antonio Torralba Barriuso
  • Patent number: 8913795
    Abstract: A method for identifying the connectivity of texture types represented in a digital image comprising pixels, each pixel having a texture value which is representative of texture at a respective position, the method comprising: partitioning the texture values into local neighborhoods of texture values; determining a directionality for each neighborhood; using the directionality of the neighborhoods to determine their nearest neighborhood or neighborhoods; connecting the neighborhoods with their nearest neighborhood or neighborhoods; and determining the connectivity of the texture types of the digital image based on the connections formed between neighborhoods.
    Type: Grant
    Filed: November 25, 2010
    Date of Patent: December 16, 2014
    Assignee: Aberystwyth University
    Inventors: Reyer Zwiggelaar, Harry Strange
  • Patent number: 8897575
    Abstract: Systems and methods for object detection are presented herein. Embodiments of the present invention utilizing a cascade feature, one or more features at different scales, one or more multi-scale features in combination with a perspective feature, or combinations thereof to detect an object of interest in an input image. In embodiments, the various features are used to train classifiers. In embodiments, the trained classifiers are used in detecting an object of interest in one or more input images.
    Type: Grant
    Filed: January 13, 2012
    Date of Patent: November 25, 2014
    Assignee: Seiko Epson Corporation
    Inventors: Yuanyuan Ding, Jing Xiao
  • Patent number: 8891879
    Abstract: An image processing apparatus for extracting a contour of an object which extracts boundary candidate points from the image and sets reference areas each having one of boundary candidate points as a center and includes small areas. A first reference area is set from the reference areas and a first small area is set from small areas included in the first reference area. From reference areas, a second reference area is set, whose feature is similar to that of the first reference area and as a second small area, a small area is set which has a feature amount similar to that of the first small area among small areas included in the second reference area. The boundary candidate points is tracked to extract the contour of the object based on the first reference area, the second reference area, the first small area, and the second small area.
    Type: Grant
    Filed: August 17, 2010
    Date of Patent: November 18, 2014
    Assignee: Canon Kabushiki Kaisha
    Inventors: Takayuki Iwamoto, Kenji Tsukamoto, Yasuo Katano
  • Patent number: 8867828
    Abstract: A method for detecting a text region in an image is disclosed. The method includes detecting a candidate text region from an input image. A set of oriented gradient images is generated from the candidate text region, and one or more detection window images of the candidate text region are captured. A sum of oriented gradients is then calculated for a region in one of the oriented gradient images. It is classified whether each detection window image contains text by comparing the associated sum of oriented gradients and a threshold. Based on the classifications of the detection window images, it is determined whether the candidate text region is a true text region.
    Type: Grant
    Filed: December 13, 2011
    Date of Patent: October 21, 2014
    Assignee: QUALCOMM Incorporated
    Inventors: Chunghoon Kim, Hyung-Il Koo, Kyu Woong Hwang
  • Patent number: 8855388
    Abstract: An analysis of a digitized image is provided. The digitized image is repeatedly convolved to form first convolved images, which first convolved images are convolved a second time to form second convolved images. Each first convolved image and the respective second convolved image representing a stage, and each stage represents a different scale or size of anomaly. As an example, the first convolution may utilize a Gaussian convolver, and the second convolution may utilize a Laplacian convolver, but other convolvers may be used. The second convolved image from a current stage and the first convolved image from a previous stage are used with a neighborhood median determined from the second convolved image from the current stage by a peak detector to detect peaks, or possible anomalies for that particular scale.
    Type: Grant
    Filed: April 29, 2011
    Date of Patent: October 7, 2014
    Assignee: vuCOMP, Inc.
    Inventors: Jeffrey C. Wehnes, James P. Monaco, David S. Harding, James H. Pike, Anbinh T. Ho, Lawrence M. Hanafy
  • Publication number: 20140254926
    Abstract: Systems, methods and computer-readable storage media are disclosed for accelerating bitmap remoting by extracting non-grid tiles from source bitmaps. A server takes a source image, identifies possibly repetitive features, and tiles the image. For each tile that contains part of a possibly repetitive feature, the server replaces that part with the dominant color of the tile. The system then sends to a client a combination of new tiles and features, and indications to tiles and features that the client has previously received and stored, along with an indication of how to recreate the image based on the tiles and features.
    Type: Application
    Filed: May 23, 2014
    Publication date: September 11, 2014
    Applicant: MICROSOFT CORPORATION
    Inventors: Nadim Y. Abdo, Voicu Anton Albu, Charles Lawrence Zitnick, III
  • Patent number: 8830398
    Abstract: It is determined whether or not an input image is an image converted from an image with a relatively low resolution based on one frame of an image. A resolution determination device includes: an edge strength calculator configured to obtain an edge strength of a pixel included in an input image based on luminance of the pixel and luminance of a pixel adjacent to the pixel, for each of a plurality of pixels included in the input image; and a resolution determiner configured to determine whether or not the input image is an image upconverted from an image with a predetermined resolution or less, based on distribution of the edge strengths.
    Type: Grant
    Filed: December 18, 2012
    Date of Patent: September 9, 2014
    Assignee: Panasonic Corporation
    Inventors: Koichi Inoue, Shinichi Tomioka, Atsuhisa Kageyama
  • Patent number: 8798377
    Abstract: Method, system and computer program for efficiently extracting and describing scale-space interest points. It is designed towards low overall computational complexity. On one hand, the data acquired during extraction in the description phase is intensively re-used. On the other hand, an algorithmic optimization of the description that dramatically speeds up the process, is proposed. First, the image is filtered with triangle kernel at different scales. The triangle filtered images are reused for extraction of the keypoints dominant orientation and the computation of the DAISY-like descriptor.
    Type: Grant
    Filed: September 1, 2010
    Date of Patent: August 5, 2014
    Assignee: Telefonica, S.A.
    Inventors: David Marimon Sanjuan, Arturo Bonnin Llofriu, Tomasz Adamek, Roger Gimeno Hernandez
  • Publication number: 20140212050
    Abstract: A method for processing an image is described. Mask bits are determined for a current pixel. The mask bits indicate intensity comparisons between the current pixel and multiple neighboring pixels. The mask bits also indicate whether each of the current pixel's neighboring pixels have been processed. A next pixel is selected for processing based on the mask bits.
    Type: Application
    Filed: July 5, 2013
    Publication date: July 31, 2014
    Inventors: Bo Zhou, Zhijie Deng, Xinping Zhang, Junchen Du, Ning Bi, Suhail Jalil