Abstract: Systems and methods to reduce the overlap areas of the fields of view of multiple cameras when reading symbols (e.g., barcode symbols, text) disposed on moving objects using an imaging system which includes an array of cameras. Each of a plurality of cameras may have a field of view directed toward objects positioned on a conveyor which moves the objects in a longitudinal direction, and the field of view of each of the plurality of cameras may overlap with the field of view of at least one other camera in the plurality of cameras in one or both of the longitudinal direction and a lateral direction orthogonal to the longitudinal direction.

Abstract: In some implementations, an image classification system of an autonomous or semi-autonomous vehicle is capable of improving multi-object classification by reducing repeated incorrect classification of objects that are considered rarely occurring objects. The system can include a common instance classifier that is trained to identify and recognize general objects (e.g., commonly occurring objects and rarely occurring objects) as belonging to specified object categories, and a rare instance classifier that is trained to compute one or more rarity scores representing likelihoods that an input image is correctly classified by the common instance classifier. The output of the rare instance classifier can be used to adjust the classification output of the common instance classifier such that the likelihood of input images being incorrectly classified is reduced.

Abstract: The present disclosure includes methods and systems for searching for digital visual media based on semantic and spatial information. In particular, one or more embodiments of the disclosed systems and methods identify digital visual media displaying targeted visual content in a targeted region based on a query term and a query area provide via a digital canvas. Specifically, the disclosed systems and methods can receive user input of a query term and a query area and provide the query term and query area to a query neural network to generate a query feature set. Moreover, the disclosed systems and methods can compare the query feature set to digital visual media feature sets. Further, based on the comparison, the disclosed systems and methods can identify digital visual media portraying targeted visual content corresponding to the query term within a targeted region corresponding to the query area.

Abstract: Analog written content of handwritten drawings or written characters can be transformed into digital ink via an analog-to-ink service. The analog-to-ink service can receive a static image of the analog written content, extract analog strokes from other information, such as background, in the static image, and then convert the analog strokes to digital ink strokes, for example, by populating an ink container with at least two parameters for defining the digital ink strokes. The at least two parameters can include a pressure, at tilt, a direction, a beginning point, an end point, a direction, a color, an order, an overlap, a language, and a time. The analog-to-ink service can provide the ink container to a content creation application that supports inking so that a user can manipulate the content in an inking environment.

Abstract: A principal-subordinate relationship between two subjects is decided with regard to multiple subjects included in an image. Similarly, with regard also to other images, a principal-subordinate relationship between two subjects is decided with regard to multiple subjects included in each image. In a case where a relationship exists in which a first subject is a subordinate subject and a second subject is principal subject, the system calculates a principal-subordinate relationship value that is higher than in a case where a relationship in which a first subject is a subordinate subject and a second subject is principal subject does not exist. The greater the number of subjects to which the first subject is subordinate from among the multiple subjects, the lower the principal-subordinate relationship value calculated. On the basis of the calculated principal-subordinate relationship values, a subject evaluation value is calculated for every subject.

Abstract: The present disclosure relates to a system, method and storage medium for generating an image. At least one processor, when executing instructions, may perform one or more of the following operations. When raw data is received, a plurality of iterations may be implemented. During each iteration, a first voxel value relating to a first voxel in an image is calculated; at least a portion of a second voxel may be continuously changed with respect to at least a portion of the first voxel value; the image may be transformed to a projection domain to generate an estimated projection based on the first voxel value and the second voxel value; a projection error may be obtained based on the estimated projection and the raw data; and the image may be corrected or updated based on the projection error.

Abstract: An electronic device for generating a corrected depth map is described. The electronic device includes a processor. The processor is configured to obtain a first depth map. The first depth map includes first depth information of a first portion of a scene sampled by the depth sensor at a first sampling. The processor is also configured to obtain a second depth map. The second depth map includes second depth information of a second portion of the scene sampled by the depth sensor at a second sampling. The processor is additionally configured to obtain displacement information indicative of a displacement of the depth sensor between the first sampling and the second sampling. The processor is also configured to generate a corrected depth map by correcting erroneous depth information based on the first depth information, the second depth information, and the displacement information.

Abstract: A banknote crack detection method and a banknote crack detection apparatus are provided. The banknote crack detection apparatus includes an image collection unit, configured to collect an optical transmission image of an inputted banknote, a pre-processing unit, configured to receive a banknote image signal collected by the image collection unit, and perform foreground region positioning and cutting on the banknote image signal; a crack detection unit, configured to detect whether a crack exists; and a determining unit, configured to determine whether the banknote is qualified according to a detection result of the crack detection unit; and a control unit, configured to control operation of each of the above units.

Abstract: A method for learning parameters of CNNs capable of identifying objectnesses by detecting bottom lines and top lines of nearest obstacles in an input image is provided. The method includes steps of: a learning device, (a) instructing a first CNN to generate first encoded feature maps and first decoded feature maps, and instructing a second CNN to generate second encoded feature maps and second decoded feature maps; (b) generating first and second obstacle segmentation results respectively representing where the bottom lines and the top lines are estimated as being located per each column, by referring to the first and the second decoded feature maps respectively; (c) estimating the objectnesses by referring to the first and the second obstacle segmentation results; (d) generating losses by referring to the objectnesses and their corresponding GTs; and (f) backpropagating the losses, to thereby learn the parameters of the CNNs.

Abstract: Biometric enrollment and verification techniques for ocular-vascular, periocular, and facial regions are described. Periocular image regions can be defined based on the dimensions of an ocular region identified in an image of a facial region. Feature descriptors can be generated for interest points in the ocular and periocular regions using a combination of patterned histogram feature descriptors. Quality metrics for the regions can be determined based on region value scores calculated based on texture surrounding the interest points. A biometric matching process for calculating a match score based on the ocular and periocular regions can progressively include additional periocular regions to obtain a greater match confidence.

Abstract: The present disclosure relates to using a neural network to efficiently denoise images that were generated by a ray tracer. The neural network can be trained using noisy images generated with noisy samples and corresponding denoised or high-sampled images (e.g., many random samples). An input feature to the neural network can include color from pixels of an image. Other input features to the neural network, which would not be known in normal image processing, can include shading normal, depth, albedo, and other characteristics available from a computer-generated scene. After the neural network is trained, a noisy image that the neural network has not seen before can have noise removed without needing manual intervention.

Abstract: On the basis of subsidiary information associated with image data, an image for the image data is segmented into multiple subregions, and feature values are extracted for each of the subregions obtained through the segmentation. The category for each of the subregions is determined on the basis of the extracted feature values.

Abstract: A method for collecting and processing remotely sensed imagery in order to achieve precise spatial co-registration (e.g., matched alignment) between multi-temporal image sets is presented. Such precise alignment or spatial co-registration of imagery can be used for change detection, image fusion, and temporal analysis/modeling. Further, images collected in this manner may be further processed in such a way that image frames or line arrays from corresponding photo stations are matched, co-aligned and if desired merged into a single image and/or subjected to the same processing sequence. A second methodology for automated detection of moving objects within a scene using a time series of remotely sensed imagery is also presented. Specialized image collection and preprocessing procedures are utilized to obtain precise spatial co-registration (image registration) between multitemporal image frame sets.

Abstract: Provided are an image evaluation system, as well as an image evaluation method, and recording medium storing an image evaluation program, in which when an image is selected from among multiple images, images are evaluated in such a manner that an image such as a scenic image or still-life image will be selected with little image bias. To achieve this, an individual image composition/subject matrix is found, the matrix comprising types of composition and types of subject included in an image. A total image composition/subject matrix is obtained by adding individual image composition/subject matrices regarding respective ones of multiple images. A composition evaluation matrix and a subject evaluation matrix are found from the total image composition/subject matrix. Image evaluation values regarding respective ones of the multiple images are calculated from a matrix product of the individual image composition/subject matrix, composition evaluation matrix and subject evaluation matrix.

Abstract: A pattern inspection method includes: scanning an inspection substrate, to be inspected, to detect a secondary electron group emitted from the inspection substrate due to irradiation with the multiple beams; correcting individually distortion of a first region image obtained from a detection signal of secondary electrons corresponding to a corresponding first region for each beam of the multiple beams; correcting distortion of a corresponding second region image corresponding to a second region larger than the first region for each of the second region images, using data of each of the first region images in which the distortion of the corresponding first region image has been corrected; and comparing an inspection image to be inspected, in which the distortion of each of the plurality of second region images has been corrected, with a reference image of a same region to output a result thereof.

Abstract: A camera system captures images from a set of cameras to generate binocular panoramic views of an environment. The cameras are oriented in the camera system to maximize the minimum number of cameras viewing a set of randomized test points. To calibrate the system, matching features between images are identified and used to estimate three-dimensional points external to the camera system. Calibration parameters are modified to improve the three-dimensional point estimates. When images are captured, a pipeline generates a depth map for each camera using reprojected views from adjacent cameras and an image pyramid that includes individual pixel depth refinement and filtering between levels of the pyramid. The images may be used generate views of the environment from different perspectives (relative to the image capture location) by generating depth surfaces corresponding to the depth maps and blending the depth surfaces.

Abstract: The present disclosure relates to a system, method and storage medium for generating an image. At least one processor, when executing instructions, may perform one or more of the following operations. When raw data is received, a plurality of iterations may be implemented. During each iteration, a first voxel value relating to a first voxel in an image is calculated; at least a portion of a second voxel may be continuously changed with respect to at least a portion of the first voxel value; the image may be transformed to a projection domain to generate an estimated projection based on the first voxel value and the second voxel value; a projection error may be obtained based on the estimated projection and the raw data; and the image may be corrected or updated based on the projection error.

Abstract: This disclosure relates to improved sketch-based image retrieval (SBIR) techniques. The SBIR techniques utilize an architecture comprising three interconnected neural networks to enable zero-shot image recognition and retrieval based on free-hand sketches. Zero-shot learning may be implemented to retrieve one or more images corresponding to the sketches without prior training on all categories of the sketches. The neural network architecture may do so, at least in part, by training encoder hashing functions to mitigate heterogeneity of sketches and images, and by applying semantic knowledge that is learned during a limited training phase to unknown categories.

Abstract: The present disclosure relates to a method and a scanner for verifying the authenticity of identity documents. The scanner comprises a slot for receiving the identity document, at least one light source and a camera or acquiring images. The method detects, by means of the camera, the presence of the identity document in the slot of the scanner and acquires three different images of the identity document when it is illuminated with white light, UV light and IR light respectively. The images are encrypted and sent to an analyze server which in response to a request returns a second message comprising a verification report including textual information extracted from the identity document. The scanner sends a first signal to an indicating means for indicating the state of the authenticity of the identity document and the textual information to a Point of Sale, POS, system in response to that the identity document has been verified.

Abstract: The present disclosure relates to methods, a system and an analyze server for verifying the authenticity of identity documents and extracting textual information therefrom. The presence of the identity document in a slot of a scanner is detected and three different images of the identity document are acquired when the identity document is illuminated with white light, UV light and IR light, respectively. The images are encrypted and sent to an analyze server which determines the authenticity of the identity and extracts textual information. The analyze server generates an encrypted verification report including textual information extracted from the identity document and sends it to the scanner, which in turn sends a first signal to an indicating device for indicating the state of the authenticity of the identity document and the textual information to a Point of Sale (POS) system in response to verifying the identity document.