Abstract: A machine learning method operates on training instances from a plurality of domains including one or more source domains and a target domain. Each training instance is represented by values for a set of features. Domain adaptation is performed using stacked marginalized denoising autoencoding (mSDA) operating on the training instances to generate a stack of domain adaptation transform layers. Each iteration of the domain adaptation includes corrupting the training instances in accord with feature corruption probabilities that are non-uniform over at least one of the set of features and the domains. A classifier is learned on the training instances transformed using the stack of domain adaptation transform layers. Thereafter, a label prediction is generated for an input instance from the target domain represented by values for the set of features by applying the classifier to the input instance transformed using the stack of domain adaptation transform domains.

Abstract: An automated system and method to authenticate one or more users based on capturing one or more images of a set of teeth, obtaining a selected image from the one or more captured images and extracting a portion of the selected image to obtain an extracted image. Each extracted image is converted into a grayscale image and stored in a database along with the username and the user keyword of the one or more users. A unique signature matrix and a pattern vector is generated by processing the grayscale image and stored in the database along with the username. One or more images comprising a set of teeth of at least one user is captured and a unique signature matrix obtained from the same is compared with a set of unique signature matrices previously stored in the database and at least one action is triggered based on the comparison.

Abstract: The purpose of the present invention is to provide a state estimation apparatus that appropriately estimates the internal state of an observation target by determining likelihoods from a plurality of observations. An observation obtaining unit of the state estimation system obtains, at given time intervals, a plurality of observation data obtained from an observable event. The observation selecting unit selects a piece of observation data from the plurality of pieces of observation data obtained by the observation obtaining unit based on a posterior probability distribution data obtained at a preceding time t?1. The likelihood obtaining unit obtains likelihood data based on the observation data selected by the observation selecting unit and predicted probability distribution data obtained through prediction processing using the posterior probability distribution data.

Abstract: A method for detecting a target object in a blind area of a vehicle is provided by embodiments of the application. The method includes: acquiring at least one frame of an image generated by shooting a blind area of a vehicle by a shooting device; extracting an image feature of an area where at least one predetermined detection gate is located in the image; and matching the image feature of the area where the predetermined detection gate is located with a matching template of the predetermined detection gate to determine a target object. An apparatus for detecting a target object in a blind area of a vehicle is further provided by embodiments of the application.

Abstract: The invention relates to a method for detecting defects in the walls of hollow glass items that rotate about their longitudinal axis during inspection, which method, in one form, is characterized by the production of an image series (4) that is binarised using a grey-scale value threshold image (5), wherein the threshold values of the threshold image (5) are measured differently according to the fixed positions of radiation-receiving elements used to display images. The images of the image series (10) formed by binarising the original image series (4) are combined in a histogram (6) that locally adds up the frequency of an occurrence of, inter alia, defect pixels representing static reflections within the sensor regions of the image series (10).

Abstract: A three-dimensional coordinate computing apparatus includes an image selecting unit and a coordinate computing unit. The image selecting unit selects a first selected image from multiple captured images, and selects a second selected image from multiple subsequent images captured by the camera after the first selected image has been captured. The second selected image is selected based on a distance between a position of capture of the first selected image and a position of capture of each of the multiple subsequent images and the number of corresponding feature points, each of which corresponds to one of feature points extracted from the first selected image and one of feature points extracted from each of the multiple subsequent images. The coordinate computing unit computes three-dimensional coordinates of the multiple corresponding feature points based on two-dimensional coordinates of each corresponding feature point in the first and second selected images.

Abstract: Methods, apparatuses, systems, and software for extended phase correction in phase sensitive Magnetic Resonance Imaging. A magnetic resonance image or images may be loaded into a memory. Two vector images A and B associated with the loaded image or images may be calculated either explicitly or implicitly so that a vector orientation by one of the two vector images at a pixel is substantially determined by a background or error phase at the pixel, and the vector orientation at the pixel by the other vector image is substantially different from that determined by the background or error phase at the pixel. A sequenced region growing phase correction algorithm may be applied to the vector images A and B to construct a new vector image V so that a vector orientation of V at each pixel is substantially determined by the background or error phase at the pixel.

Abstract: An apparatus for identifying a candidate area in a first image corresponding to an object in a second image, includes a memory and a processor to divide the plurality of candidate areas into a plurality of small candidate areas, divide an image area of the object into a plurality of small areas, perform first comparison processing for a first part, when there is a first candidate area lacking image information of the small candidate area corresponding to the first part, perform second comparison processing for a second part, predict missing result on the small candidate area corresponding to the first part in the first candidate area based on a result of the first comparison processing on a candidate area other than the first candidate area, and a result of the second comparison processing on the plurality of candidate areas, and identify the candidate area based on a prediction.

Abstract: Disclosed is a method and system for constructing a 3D structure. The system of the present disclosure comprises an image capturing unit for capturing images of an object. The system comprises of a gyroscope, a magnetometer, and an accelerometer for determining extrinsic camera parameters, wherein the extrinsic camera parameters comprise a rotation and a translation of the images. Further the system determines an internal calibration matrix once. The system uses the extrinsic camera parameters and the internal calibration matrix for determining a fundamental matrix. The system extracts features of the images for establishing point correspondences between the images. Further, the point correspondences are filtered using the fundamental matrix for generating filtered point correspondences. The filtered point correspondences are triangulated for determining 3D points representing the 3D structure. Further, the 3D structure may be optimized for eliminating reprojection errors associated with the 3D structure.

Abstract: A method to analyze an image and determine whether to output image data associated with an area of the image is provided. An object detection algorithm using training image data to detect an object based at least in part on a similarity of appearance of image data to data derived from the training image data is provided. Weakly detected objects are classified based on characteristics associated with the weakly detected object and may be added to the training image dataset for use in further training of the object detection algorithm. The object detection algorithm is trained with a revised dataset, the revised dataset being updated with data generated by the object detection algorithm.

Abstract: A method of scaling image data, includes receiving image data at a processor in at least two channels, the image data having a first resolution for the a channel and a second, lower resolution for other channels, using known pixel values of the first resolution pixels to generate an interpolation function to be applied to second resolution pixels co-located with a subset of the first resolution pixels, and applying the interpolation function to the second resolution pixels to find the unknown pixel values.

Abstract: An image processing apparatus generates, from the captured image, a plurality of images, which respectively corresponds to ranges of individual subject distances. The apparatus then applies image processing to at least one of the images in accordance with an instruction to change the shooting distance and field angle of the captured image, and generates a combined image that corresponds to the changed shooting distance and field angle. The image processing is applied to at least one of the images such that the size of a primary image in the combined image after changing the shooting distance and the field angle does not change.

Abstract: A video quality objective assessment method based on a spatiotemporal domain structure firstly combines a spatiotemporal domain gradient magnitude and color information for calculating a spatiotemporal domain local similarity, and then uses variance fusion for spatial domain fusion. The spatiotemporal domain local similarity is fused into frame-level objective quality value, and then a temporal domain fusion model is established by simulating three important global temporal effects, which are a smoothing effect, an asymmetric track effects and a recency effect, of a human visual system. Finally, the objective quality values of the distorted video sequence are obtained. By modeling the human visual temporal domain effect, the temporal domain weighting method of the present invention is able to accurately and efficiently evaluate the objective quality of the distorted video.

Abstract: The disclosure concerns processing of electronic images, such as hyperspectral or multispectral images (302) to determine an output color value for a component spectrum of a decomposed multispectral or hyperspectral image. A processor (120) first determines or accesses the component spectrum and a first color value for the component spectrum and receives an input color value for a pixel location in the image, wherein the component spectrum contributes to the pixel. The processor (120) then determines the output color value for the component spectrum based on the first color value for the component spectrum, the input color value and a utility score and stores the output color value for the component spectrum on a datastore (118). It is an advantage that the processor (120) receives a color value for a pixel location but the output color value is not exactly the received color value but one that is determined based on a utility score.

Abstract: In order to classify the presented pose of a human hand, a feature distribution map of the pose is compared to reference feature distribution maps. To generate a feature distribution map, each contour point of a hand is analyzed to determine a corresponding feature set. The feature set of a contour point includes a distance feature and an angle feature of the contour point in relation to one of its neighboring contour points. The feature sets generated from an observed pose are compared to feature sets of reference poses to determine which of the reference poses most closely matches the presented pose.

Abstract: A computer-implemented method and system register plural images using a computer processor and computer memory, where computer code stored in the computer memory causes the computer processor to perform the registration. The registration includes receiving a reference image; receiving a sensed image; computing an approximate image transformation between the reference image and the sensed image; finding plural candidate tie points by utilizing template matching; applying radiometric filtering to at least a portion of the plural candidate tie points; applying geometric filtering to at least a portion of the plural candidate tie points; computing a calculated image transformation using candidate points of the plural candidate points that were not filtered out by the radiometric and geometric filtering; transforming the sensed image using the calculated image transformation; and registering the sensed image with the reference image.

Abstract: Various embodiments provide a method for computing color descriptors of product images. For example, a number of fine color representatives can be determined to describe color variation in an image as a histogram by assigning a saturation value and a brightness value to a plurality of color hues. For each pixel of the image, the closest color among a defined fine color representative set is computed. In this example, each of the pixels is assigned a color ID corresponding to their closest matching fine color representative and at least one family color ID corresponding one or more pure color families. In this example, a histogram of the color representatives and a histogram for the color families are computed. A single color vector descriptor for the image is then determined by combining the family histogram with the color representative histogram.

Abstract: An image processing display and an image processing display program that can prevent interference with catheter treatment even when the distal end of a catheter is located behind a treatment apparatus on X-ray fluoroscopic images. The image processing display displays X-ray fluoroscopic images to be obtained by looking through a region including blood vessels using X-rays during treatment, and includes a treatment apparatus image extractor and an image processor. The treatment apparatus image extractor, based on X-ray fluoroscopic images, extracts treatment apparatus images representing the treatment apparatus having the property of absorbing X-rays. The image processor determines the distal end position of the catheter to superimpose and display images representing the determined distal end position of the catheter on the region of the X-ray fluoroscopic images showing the treatment apparatus.

Abstract: Techniques for recommending footwear based on scan data describing feet of a user and a variety of other information are described. In some instances, a footwear service may obtain scan data describing feet of a user. The footwear service may process the scan data to generate a 3D representation of the user's feet, such as a 3D model or other representation. The footwear service may also obtain other information about the user, such as user preferences, orthotics data, information identifying an activity that the user participates in and so on. The footwear service may determine a footwear recommendation for the user based on the 3D representation of the user's feet, footwear data describing a footwear item, the user preferences and/or other information.

Abstract: Provided is an image processing device including a global motion detection unit configured to detect a global motion indicating a motion of an entire image, a local motion detection unit configured to detect a local motion indicating a motion of each of areas of an image, and a main subject determination unit configured to determine a main subject based on the global motion and the local motion.