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.

Abstract: A method of obtaining contact information for a person or an entity is disclosed. The method comprises detecting a name of the person or the entity in a frame from a screen of a device; determining a presence status for the detected name based on one or more connectivity status for the detected name in one or more communication applications and/or one or more databases, where the one or more communication applications operate on the device from which the frame originates and where the one or more databases are accessible on the device; displaying the presence status on the screen of the device.

Abstract: A computerized system mountable on a moving vehicle. The computerized system includes a camera. The camera captures in real time image frames of the environment in the field of view of the camera and transfers the image frames to an image processor. The image processor is programmed for performing traffic sign recognition and for performing another driver assistance function. Information is exchanged between the traffic sign recognition and the other driver assistance function.

Abstract: Automatic roof identification systems and methods are described. Example embodiments include a roof estimation system configured to automatically detect a roof in a target image of a building having a roof. In one embodiment, automatically detecting a roof in a target image includes training one or more artificial intelligence systems to identify likely roof sections of an image. The artificial intelligence systems are trained on historical image data or an operator-specified region of interest within the target image. Then, a likely outline of the roof in the target image can be determined based on the trained artificial intelligence systems. The likely roof outline can be used to generate a roof estimate report. This abstract is provided to comply with rules requiring an abstract, and it is submitted with the intention that it will not be used to interpret or limit the scope or meaning of the claims.

Abstract: Embodiments of the present disclosure include methods, apparatuses, and systems for identifying regions to be filtered during processing of an image. A map generator is configured to receive pixels of an image and to determine a map for the image. The map indicates a number of islands within blocks of pixels of the image, where individual ones of the islands include either a single dark pixel, or two or more contiguous dark pixels, bordered by light pixels and/or one or more edges of the blocks of pixels. An index generator sets, based at least on the number of islands within the individual ones of the blocks of pixels that correspond to regions of the scanned image, filter indices for the regions of the scanned image. An image filter component, based on the filter indices, filters the regions of the scanned image.

Abstract: Systems, methods, and computer storage mediums are provided for matching multiple photographs together. An example method includes receiving a first collection of photographic images. The photographic images in the first collection are clustered into one or more composite sets of photographic images based on a comparison of the metadata associated with each photographic image in the first collection meeting a predetermined similarity threshold. An image overlap is determined between each photographic image within each of the one or composite sets of photographic images. When the image overlap exceeds a predetermined image overlap threshold, a pair of photographic images are matched, for all of the photographic images within each composite set of photographic images to form one or more composite images.

Abstract: A handwritten character input device includes a correction amount computation unit which performs, when a characteristic value of handwritten characters input to a designated character writing area of a template image by a handwriting input operation of a user does not meet predetermined character writing requirements, a computation process that computes a correction amount of the characteristic value required to meet the character writing requirements; and a corrected handwritten character information output unit which outputs information of after-correction handwritten characters generated based on the correction amount computed by the correction amount computation unit, the characteristic value of the after-correction handwritten characters meeting the character writing requirements.

Abstract: A method and system for preprocessing text containing region of a video The invention provides a method and system for preprocessing the text containing region of video for improving the optical character recognition input.

Abstract: Methods and systems for determining dance steps based on music and/or other dancers. A wearable computing system may include a head mounted display (HMD). The wearable computing system may receive a media sample including an audio sample associated with a song and/or a video sample associated with one or more dancers performing a dance. The wearable computing system may communicate the media sample to a content analysis server that may include a content identification module. The content identification module may provide information associated with a content of the media sample, such as identification of the song and the dance, to the wearable computing system. The wearable computing system may determine dance steps corresponding to the content of the media sample and may generate a display of the dance steps on the HMD.

Abstract: 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.

Abstract: Systems and methods are provided for generating a distance metric. An image manipulation application receives first and second input images. The image manipulation application generates first and second sets of points corresponding to respective edges of a first object in the first input image and a second object in the second input image. The image manipulation application determines costs of arcs connecting each point from the first set to each point of the second set based on point descriptors for each point of each arc. The image manipulation application determines a minimum set of costs between the first set and the second set that includes a cost of each arc connecting each point of the second set to a point in the first set. The image manipulation application obtains, based at least in part on the minimum set of costs, a distance metric for first and second input images.

Abstract: A technique for authenticating a user is described. During this authentication technique, an electronic device (such as a cellular telephone) captures multiple images of the user while the user moves the electronic device in a pre-defined manner (for example, along a path in 3-dimensional space), and determines positions of the electronic device when the multiple images were captured. Then, the electronic device compares the images at the positions with corresponding pre-existing images of the user captured at different points of view. If the comparisons achieve a match condition, the electronic device authenticates the user. In this way, the authentication technique may be used to prevent successful replay attacks.

Abstract: Methods and apparatuses for compressive sensing that enable efficient recovery of features in an input signal based on acquiring a few measurements corresponding to the input signal. One method of compressive sensing includes folding an image to generate first and second folds, and recovering a feature of the image based on the first and second folds without reconstructing the image. One example of a compressive sensing apparatus includes a lens, a focal plane array coupled to the lens and configured to generate first and second folds based on the image, and a decoder configured to receive the first and second folds and to recover a feature of the image without reconstructing the image. The feature may be a local geometric feature or a corner. Compressive sensing methods and apparatuses for determining translation and rotation between two images are also disclosed.

Abstract: A machine vision system for automatically identifying and inspecting objects is disclosed, including composable vision-based recognition modules and a decision algorithm to perform the final determination on object type and quality. This vision system has been used to develop a Projectile Identification System and an Automated Tactical Ammunition Classification System. The technology can be used to create numerous other inspection and automated identification systems.

Abstract: There is provided an information processing device including a database that stores feature quantities of two or more images, the database being configured such that identification information for identifying an object in each image and an attribute value related to a lighting condition under which the object was imaged are associated with a feature quantity of each image, an acquisition unit configured to acquire an input image captured by an imaging device, and a recognition unit configured to recognize an object in the input image by checking a feature quantity determined from the input image against the feature quantity of each image stored in the database. The feature quantities stored in the database include feature quantities of a plurality of images of an identical object captured under different lighting conditions.

Abstract: Faces in images are quickly detected with minimal memory resource usage. Instead of calculating a Haar-like feature value by subtracting the average pixel intensity value in one rectangular region from the average pixel intensity value in another, adjacent rectangular region, a face-detection system calculates that Haar-like feature value by dividing the average pixel intensity value in one rectangular region by the average pixel intensity value in another adjacent rectangular region. Thus, each Haar-like value is calculated as a ratio of average pixel intensity values rather than as a difference between such average pixel intensity values.

Abstract: Faces in images are quickly detected with minimal memory resource usage. Instead of calculating a Haar-like feature value by subtracting the average pixel intensity value in one rectangular region from the average pixel intensity value in another, adjacent rectangular region, a face-detection system calculates that Haar-like feature value by dividing the average pixel intensity value in one such rectangular region by the average pixel intensity value in the other such adjacent rectangular region. Thus, each Haar-like value is calculated as a ratio of average pixel intensity values rather than as a difference between such average pixel intensity values.

Abstract: The user interaction system comprises a portable pointing device (101) connected to a camera (102) and sending pictures to a digital signal processor (120), capable of recognizing an object (130) and a command given by the user (100) by moving the pointing device (101) in a specific way, and controlling an electrical apparatus (110) on the basis of this recognition.

Abstract: An improved system and method for capturing and uploading pertinent information related to a motor vehicle that is accurate, simple to use, and may be implemented on a wide-array of mobile devices in a cost-effective manner. Methods are also disclosed for users of the mobile devices to send identifying information to a database, where the identifying information is compared to other motor vehicle identifying information located the database.

Abstract: For generating a 3D geometric model (44) and/or a definition of the 3D geometric model from a single digital image of a building facade (4), a facade structure is detected from the digital image by dividing the facade (4) along horizontal lines into horizontal layers representative of floors (41), and by dividing the horizontal layers along vertical lines into tiles (42). The tiles (42) are further subdivided into a hierarchy of rectangular image regions (43). 3D architectural objects (45) corresponding to the image regions (43) are determined in an architectural element library. The 3D geometric model (44) or the definition of the 3D geometric model is generated based on the facade structure, the hierarchy and the 3D architectural objects (45). The library-based generation of the 3D geometric model makes it possible to enhance simple textured building models constructed from aerial images and/or ground-based photographs.

Abstract: A method is described for determining eye location on a frontal face digital image in order to validate the frontal face as such and determine points of reference. The method consists of the following steps: obtain a frontal face digital image that is W pixels wide and H pixels high in grayscale, where the top left vertex of the digital image is considered to be the reference source; detect each eye, starting from a search subimage of the frontal face digital image, delivering approximate coordinates if both eyes have been detected, or an arbitrary predetermined value to indicate no eye detection; locate the iris of each eye, if both eyes have been detected, starting from the approximate coordinates, validating the frontal face digital image by delivering eye positions; or, validate the frontal face digital image if one or both eyes have not been detected by submitting a search subimage defined by an area resulting from centering a square of area L in the original image, where L=0.

Abstract: An in-vivo device captures images of the GI system and transfers frames to an external system for analysis. As frames are transferred from the in-vivo device to the external system frames may be classified as belonging to a particular GI section. Based on statistical analysis of the classes, a temporary determination may be made, that the in-vivo device has transitioned to a “target section”, and, as a result of the temporary determination, the operation mode of the in-vivo device may temporarily change from a first mode to a second mode. If the temporary determination is followed by a determination which is based on analysis of additional classes, the determination that the in-vivo device is in the target section may be made final and the in-vivo device may remain in the second mode. Otherwise, the first operation mode may be resumed and the whole transition determination process may start anew.

Abstract: Described is a system for object detection using classification-based learning. A fusion method is selected, then a video sequence is processed to generate detections for each frame, wherein a detection is a representation of an object candidate. The detections are fused to generate a set of fused detections for each frame. The classification module generates a classification score labeling each fused detection based on a predetermined classification threshold. Otherwise, a token indicating that the classification module has abstained from generating a classification score is generated. The scoring module produces a confidence score for each fused detection based on a set of learned parameters from the learning module and the set of fused detections. The set of fused detections are filtered by the accept-reject module based on one of the classification score or the confidence score. Finally, a set of final detections representing an object is output.

Abstract: The present invention is related to systems and methods for translating language text on a mobile camera device offline without access to the Internet. More specifically, the present invention relates to systems and methods for displaying text of a first language and a translation of the first language text into a second language text which is displayed in real time in augmented reality on the mobile device. The processing can use a single line or a multiline algorithm designed with a plurality of processing innovations to insure accurate real-time translations without motion jitter. The invention may be used to help travelers in a foreign country with difficulties in reading and understanding text written in the local language of that country. The present invention may be utilized with wearable computers or glasses, producing seamless augmented reality foreign language translations. Some embodiments are particularly useful in translations from Asian languages to English.

Abstract: Various systems, methods, and programs embodied in computer-readable mediums are provided for the detection of patterns. In one embodiment, a pattern detection method is provided that comprises the step of performing a fractal analysis of a pattern to generate a plurality of scaling parameters from a fractal associated with the pattern in a computer system. In addition, the method further comprises the step of detecting a degree of organization in the pattern by examining a degree of equality among the scaling parameters of the fractal in the computer system.

Abstract: A system and method for creating a table from one or more form images is described. An Image Based Document Management (IBDM) server includes a stroke identification module, an image generator, a table generator and a preference module. The stroke identification module receives completed form data from one or more forms with handwriting fields, media fields and symbolic fields. The image generator generates field images that are a combination of handwriting fields, media fields and symbolic fields. The user preference module determines a user preference for displaying the table. The table generator generates a table from the field images based on the user preference.

Abstract: Disclosed are various embodiments for performing an analysis on images of a product packaging. Capture of at least one image of at least one side of a packaging in initiated. An image analysis is performed on the at least one image. Product attributes are detected based at least upon the at least one image analysis and associated with the product in a data store.

Abstract: A recording device and a control method for a recording device improve the accuracy of reading MICR information while also shortening the time required for recording media processing. A dot impact printer 10 has a magnetic head 34 that magnetically reads MICR information recorded on a recording medium S, a recording head 18 that is mounted on a different carriage than the magnetic head 34 and records images on the recording medium S, and a back scanner 112 that optically reads MICR information recorded on the recording medium S, disposed sequentially to the transportation path P of the recording medium S. When reading the MICR information by means of the magnetic head 34 does not succeed, the recording medium S is conveyed to the back scanner 112, the MICR information is read by the back scanner 112, the reading results are compared, and the MICR information is identified.

Abstract: Provided are methods for determining and analyzing photometric and morphometric features of small objects, such as cells to, for example, identify different cell states. In particularly, methods are provided for identifying apoptotic cells, and for distinguishing between cells undergoing apoptosis versus necrosis.

Abstract: Character recognition is described. In one embodiment, it may use matched sequences rather than character shape to determine a computer-legible result.

Abstract: A method to process radiological images is provided. The method comprises partitioning a radiological image of a region to be treated into a superimposition of layers, the region to be treated comprising at least one first structure and a second structure, wherein one layer solely comprises part of the first structure to be isolated from the remainder of the image, the layer solely comprising that part of the first structure to be isolated from the remainder of the image being determined by means of a parametric model of the first structure. The method further comprises determining an image of the region to be treated from the layering thus obtained, in which the isolated part of the first structure is omitted.

Abstract: Systems and methods for developing and using adaptive threshold values for different input images for object detection are disclosed. In embodiments, detector response histogram-based systems and methods train models for predicting optimal threshold values for different images. In embodiments, when training the model, an optimal threshold value for an image is defined as the value that maximizes the reduction of false positive image patches while preserving as many true positive image patches as possible. Once trained, the model may be used to set different threshold values for different images by inputting a detector response histogram for the image patches of an image into the model to determine a threshold value for detection.

Abstract: A method for determining a predictability of a media entity portion, the method includes: receiving or generating (a) reference media descriptors, and (b) probability estimations of descriptor space representatives given the reference media descriptors; wherein the descriptor space representatives are representative of a set of media entities; and calculating a predictability score of the media entity portion based on at least (a) the probability estimations of the descriptor space representatives given the reference media descriptors, and (b) relationships between the media entity portion descriptors and the descriptor space representatives. A method for processing media streams, the method may include: applying probabilistic non-parametric process on the media stream to locate media portions of interest; and generating metadata indicative of the media portions of interest.

Abstract: Disclose embodiments include an image processor for feature detection comprising a single non-planar chip containing a plurality of integrated sensing and processing resources across two or more layers adapted to capture image frames and extract image features. In a particular embodiment, the non-planar chip is a three dimensional CMOS integrated circuit (3D CMOS IC) with vertical distribution of sensing and processing resources across two or more vertical integrated circuit layers. The 3D CMOS IC implements two or more feature detectors in a single chip by reusing a plurality of circuits employed for gradient and keypoint detection. Feature detectors include a scale invariant feature transform detector (SIFT), a Harris-based feature detector, and a Hessian-based feature detector.

Abstract: A banknote detector device for an automatic teller machine, for differentiating between non-accepted and accepted banknotes, includes a banknote image sensor to receive and scan at least one face of an input banknote and to store a banknote image (BI) of each scanned. The image includes image data in the form of a number of pixels; and a reference banknote image (RBI) storage where one reference banknote image, being processed from a predetermined number of banknote images from accepted banknotes, is stored for each face of each banknote. The device includes an alignment, a banknote face classification unit, a printed pattern positioning unit and a comparison unit where, for at least one face of the banknote, the BI and RBI, being in exact pattern position in relation to each other, are compared pixel per pixel according to a predefined comparison procedure to classify the banknote as accepted or non-accepted.

Abstract: The specification and drawings present a new method, apparatus and software product for pictorial identification of a communication event using speech or text recognition in an electronic device. The communication can be (but is not limited to) a telephone call, an electronic mail message, MMS, SMS, an instant message, etc. Words from the communication event are identified using the speech or text recognition by the electronic device and at least one picture out of a library of reference pictures is identified by comparing the identified words with the key picture words using a predetermined criterion. Color background of the identified standard picture can be also identified using the identified words and a further predetermined criterion. The identified picture can be displayed during the communications event or can be stored so the user can identify the topic of the communication event later on.

Abstract: An image including a face is input (S201), a plurality of local features are detected from the input image, a region of a face in the image is specified using the plurality of detected local features (S202), and an expression of the face is determined on the basis of differences between the detection results of the local features in the region of the face and detection results which are calculated in advance as references for respective local features in the region of the face (S204).

Abstract: A method of providing image data for constructing an image of a region of a target object, comprising providing a reference diffraction pattern of a reference target object; determining an initial guess for a probe function based upon the reference diffraction pattern; and determining, by an iterative process based on the initial guess for the probe function and an initial guess for an object function, image data for a target object responsive to an intensity of radiation detected by at least one detector.

Abstract: A method includes receiving an indication of a set of image regions identified in image data. The method further includes, selecting image regions from the set of image regions for text extraction at least partially based on image region stability.

Abstract: A system is provided to correlate a medication package with a prescribed medication for a patient. The medication package accommodates an intended patient medication. The system includes an optical imager adapted to read an encoded symbol character comprising encoded patient information and further adapted to image an attribute of the medication package. The optical imager comprises a two-dimensional image sensor array and an imaging lens for focusing an image on the two-dimensional image sensor array. The two-dimensional image sensor array has a plurality of pixels formed in a plurality of rows and columns of pixels. The optical imager further includes a digital link to transmit a segment of data. The segment of data includes the patient information encoded in the encoded symbol character and the attribute of the medication package.

Abstract: An automatic inspection system is provided for monitoring rail track for detecting and generating alerts for certain hazards to the operation of trains that may cause various safety concerns and even derailment. The monitoring is during normal railway service runs without interrupting the normal railway services. In one implementation, an imaging module is used on a passenger or freight train to capture video images of the rail track under the train in motion. In one implementation, the captured video images are automatically processed in a computer using a software module which is based on digital image processing techniques and algorithms to determine whether an irregularity is present on the rail track. In one implementation, video images containing detected irregularities on the rail track are transmitted to a control center for further analysis and possible alert.

Abstract: A system and method is provided wherein, in one aspect, a processor determines whether multiple street level images have captured a nearly-identical face. If so, the images are processed to determine whether the face appears to be part of an advertisement. Once it is determined that the face is displayed on an advertisement, the boundaries of the advertisement may be determined and the location of the advertisement is stored for future use, e.g., potentially replacing the advertisement in the image with a different advertisement.

Abstract: In accordance with one embodiment, a recognition dictionary evaluation apparatus calculates, in response to the update of a recognition dictionary file, a recognition rate by item of each commodity according to feature amount data of the updated recognition dictionary file; evaluates the recognition dictionary file based on the calculated recognition rate by item of each commodity; detects, in a case in which the recognition dictionary file is evaluated to be improper, a low recognition commodity of which a correct recognition rate that a commodity is correctly recognized from all commodities is lower than a given level, and an incorrect recognition commodity of which an incorrect recognition rate is high within the recognition rate by item of the low recognition commodity; and notifies the items of the low recognition commodity and the incorrect recognition commodity.

Abstract: According to one embodiment, a device includes a search module configured to search handwriting information stored in a storage medium for a first handwriting information part whose degree of a similarity with a specific handwriting information part specified as a search key is greater than or equal to a reference value, the handwriting information includes a plurality of stroke data corresponding to a plurality of strokes, and a display processing module configured to display, on a touchscreen display, one or more thumbnail corresponding to handwriting information includes the first handwriting information part and a first user interface for changing the reference value.

Abstract: A system and method for generating a road network based on satellite imagery. Plural pixels corresponding to satellite imagery of a region are obtained. For each of the plural pixels, a probability value corresponding to the probability that the pixel belongs within the road network is calculated. A grayscale image is formed based on the calculated probability values. Plural curves are produced based on the grayscale image, wherein the producing of each curve includes positioning a shape on the grayscale image so that an average intensity of the grayscale image covered by the shape exceeds a preset threshold, moving the shape about the grayscale image while the average intensity is maintained, and tracking the movement of the shape to produce the curve. A planar-connected graph is generated by connecting at least portions of the plural curves. The planar-connected graph corresponds to the road network.

Abstract: Techniques for generating a personalized cartoon by using a few text queries are described herein. The present disclosure describes efficiently searching multiple images from a network, obtaining clipart image from the multiple images, and vectorization of the clipart image. The present disclosure also describes techniques to change a style of the cartoon such as recoloring one or more cartoon objects.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for receiving image data corresponding to an image, processing the image data to identify one or more features within the image, generating one or more keywords based on each of the one or more features, transmitting the one or more keywords to a computing device for displaying a list of the one or more keywords to a user, receiving text, the text comprising at least one keyword of the one or more keywords, that at least one keyword having been selected by the user from the list, and transmitting the image and the text for display, the text being associated with the image.

Abstract: A social photo curation system is used to automatically identify a subset of photos for an album to provide to a viewing user. The album and its photos are associated with metadata indicating information about the photos, such as individuals tagged in the photos, locations where the photos were taken, keywords or concepts associated with the photos, and the quality and variety of the photos. The social photo curation system uses this metadata to score and select the photos for a particular viewing user. The scoring and selection of photos for the album may be independent of the viewing user, or it may be customized based on the viewing user's interests and connections to other users in a social networking system.

Abstract: The invention relates to a method and an object detection device for analyzing objects in the environment and/or scenes in the environment. The object detection device includes a data processing and/or evaluation device. In the data processing and/or evaluation device, image data (xt) is evaluated on the basis of a Conditional Random Field (CRF) model and the CRF model provides additional object nodes (otn) which take into account information from an object detector.

Abstract: Unique encoding of each of a substantial number of distribution video copies of a program such as a motion picture is produced by altering the images slightly at several pre-selected locations in the program in a uniquely coded pattern. Suspected counterfeits can be compared with an unaltered master video to determine the encoded number for the copy which was counterfeited to enable tracking the source of the counterfeit. Preferably, each frame of several whole scenes is altered at each location by shifting an image so as to make the alterations largely undetectable by counterfeiters but easily detected by comparison with an unaltered master video. Artifacts are inserted in patterns representing a unique number for the program. These supplement the encoding by alteration of images and gives added means to aid in tracing counterfeit copies.