88732 results about "Pattern recognition" patented technology

A method and system operative to process color image data are disclosed. In one embodiment, the method can comprise the steps of receiving color image data, determining the color ranges to be applied to the color image data, assigning each of the pixel positions in the image data a color range, assigning a different spatial binary pattern to each color range, and assigning each of the pixel positions a binary output pixel value that corresponds to the spatial binary pattern assigned to the color range assigned to that pixel position. The resulting binary image data can be written to a file for subsequent storage, transmission, processing, or retrieval and rendering. In other embodiments, a system can be made operative to accomplish the same.

A method for determining the presence of a face from image data includes a face detection algorithm having two separate algorithmic steps: a first step of prescreening image data with a first component of the algorithm to find one or more face candidate regions of the image based on a comparison between facial shape models and facial probabilities assigned to image pixels within the region; and a second step of operating on the face candidate regions with a second component of the algorithm using a pattern matching technique to examine each face candidate region of the image and thereby confirm a facial presence in the region, whereby the combination of these components provides higher performance in terms of detection levels than either component individually. In a camera implementation, a digital camera includes an algorithm memory for storing an algorithm comprised of the aforementioned first and second components and an electronic processing section for processing the image data together with the algorithm for determining the presence of one or more faces in the scene. Facial data indicating the presence of faces may be used to control, e.g., exposure parameters of the capture of an image, or to produce processed image data that relates, e.g., color balance, to the presence of faces in the image, or the facial data may be stored together with the image data on a storage medium.

Volume dimensioning employs techniques to reduce multipath reflection or return of illumination, and hence distortion. Volume dimensioning for any given target object includes a sequence of one or more illuminations and respective detections of returned illumination. A sequence typically includes illumination with at least one initial spatial illumination pattern and with one or more refined spatial illumination patterns. Refined spatial illumination patterns are generated based on previous illumination in order to reduce distortion. The number of refined spatial illumination patterns in a sequence may be fixed, or may vary based on results of prior illumination(s) in the sequence. Refined spatial illumination patterns may avoid illuminating background areas that contribute to distortion. Sometimes, illumination with the initial spatial illumination pattern may produce sufficiently acceptable results, and refined spatial illumination patterns in the sequence omitted.

Volume dimensioning employs techniques to reduce multipath reflection or return of illumination, and hence distortion. Volume dimensioning for any given target object includes a sequence of one or more illuminations and respective detections of returned illumination. A sequence typically includes illumination with at least one initial spatial illumination pattern and with one or more refined spatial illumination patterns. Refined spatial illumination patterns are generated based on previous illumination in order to reduce distortion. The number of refined spatial illumination patterns in a sequence may be fixed, or may vary based on results of prior illumination(s) in the sequence. Refined spatial illumination patterns may avoid illuminating background areas that contribute to distortion. Sometimes, illumination with the initial spatial illumination pattern may produce sufficiently acceptable results, and refined spatial illumination patterns in the sequence omitted.

A recognition object detecting portion detects a recognition object existing in each of images which are captured in series at different time points by a camera. Images of the detected recognition object are stored in a memory. A recognition object condition judging portion determines whether image areas of the recognition object are recognizable. A movement amount detecting portion detects a movement amount of the recognition object by using the image areas of the recognition object stored in the memory. A sub-area detecting portion detects a sub-area to be recognized, by using the image areas of the recognition object read out from the memory, in accordance with the movement amount of the recognition object. A recognition portion performs recognition processing on the sub-areas. A recognition result integration portion integrates recognition results of the sub-areas including recognized characters and / or recognized patterns and the corresponding reliabilities, respectively.

The present invention relates to an apparatus for rapidly analyzing frame(s) of digitized video data which may include objects of interest randomly distributed throughout the video data and wherein said objects are susceptible to detection, classification, and ultimately identification by filtering said video data for certain differentiable characteristics of said objects. The present invention may be practiced on pre-existing sequences of image data or may be integrated into an imaging device for real-time, dynamic, object identification, classification, logging / counting, cataloging, retention (with links to stored bitmaps of said object), retrieval, and the like. The present invention readily lends itself to the problem of automatic and semi-automatic cataloging of vast numbers of objects such as traffic control signs and utility poles disposed in myriad settings. When used in conjunction with navigational or positional inputs, such as GPS, an output from the inventative systems indicates the identity of each object, calculates object location, classifies each object by type, extracts legible text appearing on a surface of the object (if any), and stores a visual representation of the object in a form dictated by the end user / operator of the system. The output lends itself to examination and extraction of scene detail, which cannot practically be successfully accomplished with just human viewers operating video equipment, although human intervention can still be used to help judge and confirm a variety of classifications of certain instances and for types of identified objects.

A system and method for improving the coding efficiency of motion vector information in video coding. According to various embodiments, a list of motion vector predictor candidates is arranged according to predefined rules. Each motion vector also has a reference index associated with it. One of the motion vector candidates is then selected as a predictor based on predefined rules, or the selection is explicitly signaled in the bitstream. The reference index associated with the selected motion vector is used as a reference index for the current block. The reference index is predicted along with the motion vector. Such embodiments can improve the compression efficiency of modern video codecs.

The invention aligns a set of plural images to construct a mosaic image. At least different pairs of the images overlap partially (or fully), and typically are images captured by a camera looking at the same scene but oriented at different angles from approximately the same location or similar locations. In order to align one of the images with another one of the images, the following steps are carried out: (a) determining a difference error between the one image and the other image; (b) computing an incremental rotation of the one image relative to a 3-dimensional coordinate system through an incremental angle which tends to reduce the difference error; and (c) rotating the one image in accordance with the incremental rotation to produce an incrementally warped version of the one image. As long as the difference error remains significant, the method continues by re-performing the foregoing determining, computing and rotating steps but this time with the incrementally warped version of the one image.

In some arrangements, product packaging is digitally watermarked over most of its extent to facilitate high-throughput item identification at retail checkouts. Imagery captured by conventional or plenoptic cameras can be processed (e.g., by GPUs) to derive several different perspective-transformed views—further minimizing the need to manually reposition items for identification. Crinkles and other deformations in product packaging can be optically sensed, allowing such surfaces to be virtually flattened to aid identification. Piles of items can be 3D-modelled and virtually segmented into geometric primitives to aid identification, and to discover locations of obscured items. Other data (e.g., including data from sensors in aisles, shelves and carts, and gaze tracking for clues about visual saliency) can be used in assessing identification hypotheses about an item. A great variety of other features and arrangements are also detailed.

Image analysis techniques, including object recognition analysis, are applied to images obtained by one or more image capture devices deployed within inventory environments. The object recognition analysis provides object recognition data (that may include one or more recognized product instances) based on stored product (training) images. In turn, a variety of functionalities may be enabled based on the object recognition data. For example, a planogram may be extracted and compared to a target planogram, or at least one product display parameter for a product can be determined and used to assess presence of the product within the inventory environment, or to determine compliance of display of the product with a promotional objective. In yet another embodiment, comparisons may be made within a single image or between multiple images over time to detect potential conditions requiring response. In this manner, efficiency and effectiveness of many previously manually-implemented tasks may be improved.

A gesture recognition apparatus for recognizing postures or gestures of an object person based on images of the object person captured by cameras. The gesture recognition apparatus includes: a face / fingertip position detection means which detects a face position and a fingertip position of the object person in three-dimensional space based on contour information and human skin region information of the object person to be produced by the images captured; and a posture / gesture recognition means which operates to detect changes of the fingertip position by a predetermined method, to process the detected results by a previously stored method, to determine a posture or a gesture of the object person, and to recognize a posture or a gesture of the object person.

A technique for hiding of data, including watermarks, in human-perceptible images, that is, image host data, is disclosed. In one embodiment a method comprises three steps. In the first step, data to be embedded is inputted. In the case of a watermark, this data is a unique signature, and may be a pseudo-noise (PN) code. In the case of hidden data to be embedded in the host data, this data is the hidden data itself, or the hidden data as spread against the frequency spectrum by a pseudo-noise (PN) code. In the second step, the inputted data is embedded within the host data, in accordance with a perceptual mask of the host data. The perceptual mask determines the optimal locations within the host data to insert the inputted data. In the case of images, these optimal locations are determined by reference to the human visual system. In the third step, the host data, with the embedded data, is further masked by a non-frequency mask. In the case of image data, the non-frequency mask is a spatial mask.

Apparatus for operating home appliances using gesture recognition are disclosed. An image receiver receives a continuous stream of images of a gesture made within the image receiver's field of view. An image processor is connected to the image receiver for sampling the received stream of gesture images to form a gesture image sequence, and for recognizing whether the gesture image sequence corresponds to one of a predefined set of gestures. If the input gesture is recognized as being one of the predefined gestures, an operations processor, which is connected to the image processor, identifies a home appliance operation associated with the recognized gesture. An appliance controller is connected to the operations processor for causing the predefined home appliance operation to be performed. Home appliances and methods for operating home appliances using gesture recognition are also disclosed.

The present invention discloses methods of digitally converting 2D motion pictures or any other 2D image sequences to stereoscopic 3D image data for 3D exhibition. In one embodiment, various types of image data cues can be collected from 2D source images by various methods and then used for producing two distinct stereoscopic 3D views. Embodiments of the disclosed methods can be implemented within a highly efficient system comprising both software and computing hardware. The architectural model of some embodiments of the system is equally applicable to a wide range of conversion, re-mastering and visual enhancement applications for motion pictures and other image sequences, including converting a 2D motion picture or a 2D image sequence to 3D, re-mastering a motion picture or a video sequence to a different frame rate, enhancing the quality of a motion picture or other image sequences, or other conversions that facilitate further improvement in visual image quality within a projector to produce the enhanced images.

Method and system for measuring physiological parameters. The method includes capturing a sequence of images of a human face and identifying the location of the face in a frame of the video and establishing a region of interest including the face. Pixels are separated in the region of interest in a frame into at least two channel values forming raw traces over time. The raw traces are decomposed into at least two independent source signals. At least one of the source signals is processed to obtain a physiological parameter.

Systems and methods may calibrate an indicator of speed of an autonomous vehicle. In one implementation, a system may include at least one processor programmed to: receive from a camera at least a plurality of images representative of an environment of the vehicle; analyze the plurality of images to identify at least two recognized landmarks; determine, based on known locations of the two recognized landmarks, a value indicative of a distance between the at least two recognized landmarks; determine, based on an output of at least one sensor associated with the autonomous vehicle, a measured distance between the at least two landmarks; determine a correction factor for the at least one sensor based on a comparison of the value indicative of the distance between the at least to recognized landmarks and the measured distance between the at least two landmarks.

A method for generating a photolithography mask for optically transferring a pattern formed in the mask onto a substrate utilizing an imaging system. The method includes the steps of: (a) defining a set of calibration patterns, which are represented in a data format; (b) printing the calibration patterns on a substrate utilizing the given imaging system; (c) determining a first set of contour patterns corresponding to the calibration patterns imaged on the substrate; (d) generating a system pseudo-intensity function, which approximates the imaging performance of the imaging system; (e) determining a second set of contour patterns by utilizing the system pseudo-intensity function to define how the calibration patterns will be imaged in the substrate; (f) comparing the first set of contour patterns and the second set of contour patterns to determine the difference therebetween; (g) adjusting the system pseudo-intensity function until the difference between the first set of contour patterns and the second set of contour patterns is below a predefined criteria; and (h) utilizing the adjusted system pseudo-intensity function to modify the mask so as to provide for optical proximity correction.

Pulse oximetry is improved through classification of plethysmographic signals by processing the plethysmographic signals using a neural network that receives input coefficients from multiple signal domains including, for example, spectral, bispectral, cepstral and Wavelet filtered signal domains. In one embodiment, a plethysmographic signal obtained from a patient is transformed (240) from a first domain to a plurality of different signal domains (242, 243, 244, 245) to obtain a corresponding plurality of transformed plethysmographic signals. A plurality of sets of coefficients derived from the transformed plethysmographic signals are selected and directed to an input layer (251) of a neural network (250). The plethysmographic signal is classified by an output layer (253) of the neural network (250) that is connected to the input layer (251) by one or more hidden layers (252).

A system and method are disclosed for generating an animatable object. A skeleton of the desired character is constructed by the user utilizing various predetermined components. These predetermined components include a various selection of rods and joints. The rods are static components which remain rigid during motion, while the various joints are moveable components. A static digitized image, for example, an image of the user, is utilized and a constructed skeleton is superimposed onto it. The desired object, such as the image of the user, can then be extracted from the background of the digital image and the resulting personal character can then be animated, for instance by selecting and dragging one of the hands with a mouse.

A technique for processing a digital image uses face detection to achieve one or more desired image processing parameters. A group of pixels is identified that corresponds to a face image within the digital image. A skin tone is detected for the face image by determining one or more default color or tonal values, or combinations thereof, for the group of pixels. Values of one or more parameters are adjusted for the group of pixels that correspond to the face image based on the detected skin tone.

Disclosed is a method for linguistic pattern recognition of information. Initially, textual information is retrieved from a data source utilizing a network. The textual information is then segmented into a plurality of phrases, which are then scanned for patterns of interest. For each pattern of interest found a corresponding event structure is built. Event structures that provide information about essentially the same incident are then merged.

This invention relates to intelligent video surveillance fuzzy logic neural networks, camera systems with local and network-shared communications for facial, physical condition and intoxication recognition. The device we reveal helps reduce underage drinking by detecting and refusing entrance or service to subjects under legal drinking age. The device we reveal can estimate attention of viewers of advertising, entertainment, displays and the like. The invention also relates to method, and Vision, Image and related-data, database-systems to reduce the volume of surveillance data through automatically recognizing and recording only occurrences of exceptions and elimination of non-events thereby achieving a reduction factor of up to 60,000. This invention permits members of the LastCall™ Network to share their databases of the facial recognition and identification of subjects recorded in the exception occurrences with participating members' databases: locally, citywide, nationally and internationally, depending upon level of sharing permission.

Image analysis techniques, including object recognition analysis, are applied to images obtained by one or more image capture devices deployed within inventory environments. The object recognition analysis provides object recognition data (that may include one or more recognized product instances) based on stored product (training) images. In turn, a variety of functionalities may be enabled based on the object recognition data. For example, a planogram may be extracted and compared to a target planogram, or at least one product display parameter for a product can be determined and used to assess presence of the product within the inventory environment, or to determine compliance of display of the product with a promotional objective. In yet another embodiment, comparisons may be made within a single image or between multiple images over time to detect potential conditions requiring response. In this manner, efficiency and effectiveness of many previously manually-implemented tasks may be improved.

A system and method for capturing image data captures images responsive to programmed sequences. The sequences are performed on a periodic basis as well as in response to inputs corresponding to alarm conditions and transactions conducted at automated banking machines or other devices. Image data may also be captured in response to image conditions including the sensing of motion or the loss of usable video from selected cameras. Image data is stored in connection with data corresponding to circumstances associated with each triggering event. Stored image data may be searched by one or more parameters. Parameters include data stored in association with each image, types of events causing image data to be stored, as well as other image conditions in stored images.