A user device detects, in a field of view of the camera, a first side of a document, and determines first information associated with the first side of the document. The user device selects a first image resolution based on the first information and captures, by the camera, a first image of the first side of the document according to the first image resolution. The user device detects, in the field of view of the camera, a second side of the document, and determines second information associated with the second side of the document. The user device selects a second image resolution based on the second information, and captures, by the camera, a second image of the second side of the document according to the second image resolution. The user device performs an action related to the first image and the second image.

A search area is set on a recognition target image, cutout areas are set at a plurality of positions in the search area, images corresponding to the plurality of set cutout areas are extracted, similarities of candidate characters obtained by comparison between the extracted images and dictionary data is weighted in accordance with the positions of the cutout areas. In such a manner, evaluation values of the candidate characters are obtained, and a candidate character with the highest evaluation value among the obtained candidate characters is output as a recognition result. Further, a search area relating to a next character is set based on position information about the cutout area corresponding to the recognition result.

In an image processing apparatus, a fold determining unit is configured to determine that the document is folded if it is determined that a number of one or more character strings included in a first group is equal to or larger than a first threshold and if a number of one or more character strings included in a second group is equal to or larger than a second threshold, the first group including multiple character strings including the first character string, which are arrayed in series and have an inclination included in the first inclination interval, the second group including multiple character strings including the second character string, which are arrayed in series and have an inclination included in the second inclination interval.

An image correction device includes a line segment detection module, a shape specification module and an image correction module. The line segment detection module detects from a captured image obtained by photographing a document a plurality of line segments that correspond to the notation on the surface of the document. The shape specification module specifies shape approximation lines that approximate the surface shape of the document from the plurality of line segments. The image correction module utilizes the shape approximation lines specified by the shape specification module to correct the captured image.

Devices and a method are provided for providing feedback to a user. In one implementation, the method comprises obtaining a plurality of images from an image sensor. The image sensor is configured to be positioned for movement with the user's head. The method further comprises monitoring the images, and determining whether relative motion occurs between a first portion of a scene captured in the plurality of images and other portions of the scene captured in the plurality of images. If the first portion of the scene moves less than at least one other portion of the scene, the method comprises obtaining contextual information from the first portion of the scene. The method further comprises providing the feedback to the user based on at least part of the contextual information.

Extracting financial card information with relaxed alignment comprises a method to receive an image of a card, determine one or more edge finder zones in locations of the image, and identify lines in the one or more edge finder zones. The method further identifies one or more quadrilaterals formed by intersections of extrapolations of the identified lines, determines an aspect ratio of the one or more quadrilateral, and compares the determined aspect ratios of the quadrilateral to an expected aspect ratio. The method then identifies a quadrilateral that matches the expected aspect ratio and performs an optical character recognition algorithm on the rectified model. A similar method is performed on multiple cards in an image. The results of the analysis of each of the cards are compared to improve accuracy of the data.

Devices and methods are disclosed generating input. In one implementation, a stylus is provided for generating writing input. The stylus includes an elongated body having a distal end, and a light source configured to project coherent light on an opposing surface adjacent the distal end. The stylus further includes at least one sensor configured to measure first reflections of the coherent light from the opposing surface while the distal end moves in contact with the opposing surface, and to measure second reflections of the coherent light from the opposing surface while the distal end moves above the opposing surface and out of contact with the opposing surface. The stylus also includes at least one processor configured to receive input from the at least one sensor and to enable determining three dimensional positions of the distal end based on the first reflections and the second reflections.

A method and system for analyzing text in an image is disclosed. A text localization and classification system accesses an annotated image comprising a plurality of text location identifiers for a given item of text. A neural network predicts the location of the given item of text using at least a first location identifier and a second location identifier. Optionally, the first location identifier comprises a first shape and the second location identifier comprises a second shape. A first loss is generated using a first loss function, the first loss corresponding to the predicated location using the first location identifier. A second loss is generated using a second loss function, the second loss corresponding to the predicated location using the second location identifier. The neural network is enhanced with backpropagation using the first loss and the second loss.

Devices and methods are disclosed for generating input. In one implementation, a stylus is provided for generating writing input. The stylus includes an elongated body having a distal end, and a light source configured to project coherent light on an opposing surface adjacent the distal end. The stylus further includes at least one sensor configured to measure first reflections of the coherent light from the opposing surface while the distal end moves in contact with the opposing surface, and to measure second reflections of the coherent light from the opposing surface while the distal end moves above the opposing surface and out of contact with the opposing surface. The stylus also includes at least one processor configured to receive input from the at least one sensor and to enable determining three dimensional positions of the distal end based on the first reflections and the second reflections.

Biometric data are obtained from biometric sensors on a stand-alone computing device, which may contain an ASIC, connected to or incorporated within it. The computing device and ASIC, in combination or individually, capture biometric samples, extract biometric features and match them to one or more locally stored, encrypted templates. The biometric matching may be enhanced by the use of an entered PIN. The biometric templates and other sensitive data at rest are encrypted using hardware elements of the computing device and ASIC, and/or a PIN hash. A stored obfuscated Password is de-obfuscated and may be released to the authentication mechanism in response to successfully decrypted templates and matching biometric samples. A different de-obfuscated password may be released to authenticate the user to a remote or local computer and to encrypt data in transit. This eliminates the need for the user to remember and enter complex passwords on the device.