Abstract: Systems and methods for controlling a vehicle are provided. The systems and methods receive a frame of visible image data from at least one camera of the vehicle and a frame of invisible perception data from a perception sensor. The systems and methods fuse the frame of invisible perception data and the frame of visible image data to provide a fused frame of perception data. Object detection, classification and tracking are performed using a neural network based on the fused frame of perception data to provide object detection data. Steering, propulsion and/or braking of the vehicle are controlled based on the object detection data.

Abstract: A flame detecting arrangement including: at least one flame detector, arranged with its field-of-view covering a predetermined area; at least one movement sensor, arranged together with the at least one flame detector; and at least one processing device. The at least one processing device is arranged to: receive signals from the movement sensor; detect movement of the flame detector based on these signals; and alert an operator and/or a control system if movement of the flame detector has been detected, in order to enable correction of the field-of-view of the flame detector to the desired field-of-view.

Abstract: Systems, apparatuses, methods, and computer program products are disclosed for generating one or more candidate event material sets. An example method includes receiving an event material generation request. The example method further includes generating one or more candidate event feature sets and one or more candidate event material sets based on the one or more candidate event feature sets. The method further includes providing the one or more candidate event material sets.

Abstract: A processing device receives security data from a plurality of web services associated with an organization and stores the security data separately in an unstructured data storage. The processing device generates one or more purpose built databases from the security data in the unstructured data storage, the one or more purpose built databases merging the security data from the plurality of web services. The processing device further receives, from a requestor, an analysis request pertaining to the plurality of web services, executes an analysis using the one or more purpose built databases to generate a response to the analysis request, and provides the response to the analysis request to the requestor.

Abstract: Embodiments of the disclosure are drawn to apparatuses, systems, methods for an internet of things (IoT) system to include edge devices that perform at least some functions without communicating with a cloud computing system. An edge device may include a memory with on-memory pattern matching capabilities. The edge device may perform pattern matching operations on data collected by the edge device or sensors in communication with the edge device. Based on results of the pattern matching operations, the edge device may perform various functions, such as transmitting data to the cloud computing system, activating an alarm, and/or changing a frequency at which data is transmitted.

Abstract: Each input image from a plurality of input images is divided into a plurality of image tiles. A feature point map including a plurality of feature point tiles respectively corresponding to the plurality of image tiles is generated by extracting feature points included in each image tile of the plurality of image tiles. A descriptor map including a plurality of descriptor tiles respectively corresponding to the plurality of feature point tiles is generated by generating descriptors of feature points included in the feature point map. Mapping information containing matching relationships between feature points included in different input images of the plurality of input images is generated based on a plurality of descriptor maps respectively corresponding to the plurality of input images. Image merging performance may be enhanced by dividing the input image into the plurality of image tiles to increase distribution uniformity of the feature points.

Abstract: A goods sensing system includes: a sample collector that collects a plurality of sets of image samples, where each set of the image samples comprise sample images of a type of goods at multiple angles, where a set of the image samples of a same type of goods are provided with a same group identification, and the group identification is the type of the goods corresponding to the set of image samples; a model trainer that trains a convolutional neural network model according to each sample image and a group identification of the sample image to obtain a goods identification model; a real-time image collector that continuously acquires at least one real-time image of space in front of a shelf, each real-time image including part or all of images of goods; and a goods category deriver that obtains a type and quantity of the goods displayed in the real-time.

Abstract: A method is provided for verifying the authenticity of an article which bears a security mark. The method includes irradiating the security mark with a time-varying light source, ascertaining at least one portion of the emissions spectrum of the irradiated security mark with at least one photodetector, determining the photoluminescence lifetime of the security mark by monitoring the time or frequency response of the photodetector, and verifying the authenticity of the article only if the security mark exhibits a photoluminescence which has a lifetime that falls within the range of appropriate values for each portion of the photoluminescence spectrum for which the photoluminescence lifetime of said security mark was ascertained.

Abstract: A method for processing images includes recognizing a hair region in a target image; acquiring a hairline direction of each pixel point in the hair region; acquiring an intermediate image by smoothing the hair region based on the hairline direction; and acquiring a virtual hairline along the hairline direction, and acquiring a first image by superimposing the virtual hairline to the hair region in the intermediate image.

Abstract: A method for use in performing localisation in a three-dimensional (3D) environment, the method including in one or more electronic processing devices: determining from a local scan performed by at least one laser sensor, intensity data based at least in part on a power of radiation returned to the at least one laser sensor from points in a local point cloud obtained from the local scan; calculating a first intensity descriptor for the local point cloud using the intensity data; retrieving a plurality of previously calculated second intensity descriptors that are each associated with a respective portion of a map of the 3D environment; comparing the first intensity descriptor with at least some of the second intensity descriptors; and, determining a location with respect to the map at least in part in accordance with results of the comparison.

Abstract: Systems and methods relate to auto-tagging of data in a data lake or a data storage. Generating a statistical summary of the data lake and interactively receiving data in a selected column of an exemplar data addresses an issue of efficiently and accurately auto-tagging data in a data lake. The present disclosure automatically generates a statistical summary of the data lake using a lightweight off-line processing. A graphical user interface interactively receives an exemplar data file with a selection of a column in the exemplar data file. A list of candidate data-tagging patterns is generated based on the statistical summary and updates the list by removing candidate data-tagging patterns that under-generalize the data. The present disclosure determines a data-tagging pattern by selecting a candidate data-tagging profile from the list based on having the least number of matching columns in the data lake.

Abstract: There are provided an information suggestion system, an information suggestion method, a program, and a recording medium capable of suggesting, to a user, information on a product matching a hobby and the preference of the user from an image group of the user. In the information suggestion system, the information suggestion method, the program, and the recording medium, an image group acquisition unit acquires an image group of a user, and an image analysis unit detects an attribute of each image. An imaging number count unit counts an imaging number of same-attribute images, and an imaging frequency calculation unit calculates an imaging frequency of the same-attribute images. An imaging purpose estimation unit estimates an imaging purpose of the same-attribute images.

Abstract: A component orientation determination data creation device for creating component orientation determination data used in a component orientation determination system that determines an orientation of the component by comparing a brightness value of a determination region specified in the component orientation determination data within an image of the component captured by a camera with a determination threshold. A difference image between an image of the component in a correct orientation and an image of the component in another orientation is calculated by acquiring multiple images of the component in different orientations by changing the orientation with respect to the camera of the component that is the target for the component orientation determination data creation.

Abstract: In implementations of segmenting objects in vector graphics images, an object segmentation system can obtain points that identify an object in a vector graphics image, and determine a region of interest in the image that includes the object based on the points that identify the object. The object segmentation system can generate a heat map from the points that identify the object in the image, and a rasterized region from rasterizing the region of interest. The object segmentation system can generate a mask from the rasterized region and the heat map, the mask identifying pixels of the object in the rasterized region, and determine, from the mask, paths of the vector graphics corresponding to the object.

Abstract: The present disclosure relates to encoding and decoding video by employing texture coding. In particular, a texture region is identified within a video picture and a texture patch is determined for the texture region. Moreover, a first set of parameters is derived specifying weighting factors for reconstructing spectral coefficients of the texture region by fitting the texture region in a spectral domain to a first function of the texture patch, the first function being defined by the first set of parameters. The texture patch and the first set of parameters are then included into a bitstream which is output by an encoder and provided in this way to the decoder which reconstructs the texture based on the patch and the function applied to the patch.

Abstract: Example aspects of the present disclosure are directed to systems and methods for learning data augmentation strategies for improved object detection model performance. In particular, example aspects of the present disclosure are directed to iterative reinforcement learning approaches in which, at each of a plurality of iterations, a controller model selects a series of one or more augmentation operations to be applied to training images to generate augmented images. For example, the controller model can select the augmentation operations from a defined search space of available operations which can, for example, include operations that augment the training image without modification of the locations of a target object and corresponding bounding shape within the image and/or operations that do modify the locations of the target object and bounding shape within the training image.

Abstract: Described herein are systems and methods that search videos and other media content to identify items, objects, faces, or other entities within the media content. Detectors identify objects within media content by, for instance, detecting a predetermined set of visual features corresponding to the objects. Detectors configured to identify an object can be trained using a machine learned model (e.g., a convolutional neural network) as applied to a set of example media content items that include the object. The systems provide user interfaces that allow users to review search results, pinpoint relevant portions of media content items where the identified objects are determined to be present, review detector performance and retrain detectors, providing search result feedback, and/or reviewing video monitoring results and analytics.

Abstract: The present disclosure describes methods and systems for handling personally identifiable information. In some implementations, the information is collected during the authentication of identification (ID) documents. The system includes a one-way hashing function that converts unique personal identifiable information into a unique digest. The system can store the digest without having to store the personal identifiable information. Because the hashing function generates the same digest when given the same input, the digests can be used as anonymized identifiers in place of the personal identifiable information.

Abstract: Systems and methods are configured to extract text from images (e.g., document images) utilizing a combination of optical character recognition processes and neural network-based analysis of various images to train a machine-learning object recognition model that is configured to identify text within images based on object-comparisons between known and unknown text within images. In certain embodiments, identified text within images may be utilized to identify corresponding regions-of-interest for extraction of image data encompassing a portion of an image that may be indexed based at least in part on text identified as corresponding to the particular region-of-interest.

Abstract: The present disclosure relates to using end-to-end differentiable pipeline for optimizing parameters of a base procedural material to generate a procedural material corresponding to a target physical material. For example, the disclosed systems can receive a digital image of a target physical material. In response, the disclosed systems can retrieve a differentiable procedural material for use as a base procedural material in response. The disclosed systems can compare a digital image of the base procedural material with the digital image of the target physical material using a loss function, such as a style loss function that compares visual appearance. Based on the determined loss, the disclosed systems can modify the parameters of the base procedural material to determine procedural material parameters for the target physical material. The disclosed systems can generate a procedural material corresponding to the base procedural material using the determined procedural material parameters.

Abstract: Described herein is a method for detecting keypoints in three-dimensional images in which a three-dimensional image of a scene captured by a depth sensing imaging system is processed using a distance-independent keypoint filter. Keypoints are derived from the three-dimensional image by determining a mean shift field and using x- and y-components of the mean shift field to derive intersections of 0-isolines thereof. Positive and negative keypoints or nodes are connected to one another, positive to positive and negative to negative, to form a keygraph structure.

Abstract: To provide an information processing system, an information processing method, and a recording medium, which are capable of assisting in a search for a moving image using a graph of data which is not obtained from image analysis associated with a moving image. Provided is an information processing system, including: a moving image data acquiring unit configured to acquire moving image data; a communication unit configured to receive sensor data associated with the moving image data and chronological data corresponding to a shooting time of the moving image data; an image signal processing unit configured to perform image analysis on the moving image data and generate image analysis result data; and a control unit configured to generate an interface including the moving image data and graphs of at least two pieces of data among the sensor data, the chronological data, and the image analysis result data.

Abstract: An anomaly detection apparatus 100 includes an image transformation unit 103 that calculates an image transformation parameter, based on an inspection image in which an inspection object appears, a reference image indicating a normal state of the inspection object and a parameter for image transformation parameter calculation, and performs image transformation on the inspection image using the image transformation parameter, an image change detection unit 104 that collates the reference image and the image-transformed inspection image using a change detection parameter, and calculates an anomaly certainty factor indicating whether there is a change in a specific region of the inspection image, a change detection parameter learning unit 106 that learns the change detection parameter, based on a difference between a training image indicating a correct answer value of the change and the anomaly certainty factor, and an image transformation parameter learning unit 108 that learns the parameter for image transform

Abstract: The present disclosure relates to a method of determining that a finger is in stable contact with a surface covering a sensor area of a fingerprint sensor. The method comprises, by means of the fingerprint sensor, acquiring a time-sequence of images of the fingerprint of the received finger. The method also comprises, for each image of the time-sequence, dividing an image area of the image, corresponding to the sensor area of the fingerprint sensor, into a plurality of image regions (r), said regions partly overlapping each other and covering the whole image area. The method also comprises, based on image analysis of each of the plurality of image regions of each image of the time-sequence, determining that the finger is in stable contact with the surface covering the sensor area.

Abstract: This invention provides an apparatus having an image capturing unit, a transforming unit which transforms RAW image obtained by the image capturing unit into a plurality of channels, and an encoder which performs a frequency transformation, quantization, and encoding for each channel, where the apparatus comprises a classifying unit which, based on a sub-band of a low-frequency component obtained by the frequency transformation for a channel, in which a luminance component is predominant, obtained by the transforming unit, classifies a luminance of each region in a case where one screen is divided into a plurality of regions; and an estimation unit which, for each luminance, computes an index value representing a degree of difference between data before/after the quantization for each region, and estimates an image quality based on the index value.

Abstract: Systems and methods for ranking one or more products in online shopping. One or more products are identified based on a search query received from user. The one or more products are ranked based on terms present in the search query. Each of the one or more products has one or more attributes associated with it. An attribute score for each of the one or more products is determined. Further, based on the attribute score, the relevancy of the one or more products is determined. Based on the relevancy, a marginal relevancy score for each of the one or more products is determined. The one or more products are re-ranked based on the marginal relevancy score. The rank of the one or more products can also be modified to optimize revenue generation.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for liveness detection are provided. One of methods includes: displaying an image to a user, and capturing a face image of the user while displaying the image to the user; determining an eye image of an eye of the user based on the face image; extracting a to-be-verified image from the eye image of the user, wherein the to-be-verified image is reflection of the displayed image in the eye of the user; comparing the displayed image with the to-be-verified image to determine whether the to-be-verified image matches the displayed image; and performing liveness detection on the user based on a result of comparison.

Abstract: Content can be located for items that are stylistically similar to an item of interest. The item of interest can be represented in a query image, which is analyzed to determine one or more regions having an item represented therein. The classification of the item is determined, enabling identification of a trained model to be used to process image data for the region(s) of the query image. The trained model outputs a set of attributes, relating to visual or stylistic attributes, and corresponding confidence or prominence values for the attributes. These attributes and values can be compared against a data repository to locate items determined to be similar based on corresponding attributes and values. A similarity determination algorithm can identify similar items and rank those items by similarity. Content for the most similar items is returned as a result for the query image.

Abstract: A computing device is described which has a memory holding at least two input images depicting different parts of a panoramic scene, the images having been captured by a user moving the camera by hand to capture the panorama. The computing device has an image stitching component configured to identify, at a processor, a region of overlap between the at least two images and to calculate a displacement vector for each of a plurality of warp points in the region of overlap. The image stitching component is arranged to warp a second one of the at least two images using the warp points; and to join the warped second image to the first image.

Abstract: Methods and systems for incorporating physical documents into a document review workflow involving electronic documents. One or more embodiments detect a presence of a physical document within a field of view of an AR device and map the physical document to an existing electronic document based on visual features of the physical document. Additionally, one or more embodiments determine at least one difference between the physical document and the electronic document and create, for the physical document and the electronic document a shared state mapping including the difference(s). One or more embodiments then apply the difference to the physical document or the electronic document by displaying the difference(s) in an AR layer within the field of view of the AR device or storing the difference(s) in the electronic document.

Abstract: A system capable of determining which recognition algorithms should be applied to regions of interest within digital representations is presented. A preprocessing module utilizes one or more feature identification algorithms to determine regions of interest based on feature density. The preprocessing modules leverages the feature density signature for each region to determine which of a plurality of diverse recognition modules should operate on the region of interest. A specific embodiment that focuses on structured documents is also presented. Further, the disclosed approach can be enhanced by addition of an object classifier that classifies types of objects found in the regions of interest.

Abstract: A fingerprint verification method and a fingerprint verification apparatus performing the fingerprint verification method are disclosed. The fingerprint verification apparatus determines a first similarity between a query fingerprint image and each of registered fingerprint images, selects a target registered fingerprint image group from registered fingerprint image groups based on the first similarity, determines a second similarity between the query fingerprint image and each of registered fingerprint images in the target registered fingerprint image group based on matching relationship information between the registered fingerprint images in the target registered fingerprint image group, and determines whether fingerprint verification of the query fingerprint image is successful based on the second similarity.

Abstract: An outside recognition device according to the present invention has an object to perform distance measurement of an obstacle detected with a surrounding monitoring camera at a low cost. The outside recognition device according to the present invention has a stereo camera that measures a distance in front of the vehicle, a surrounding monitoring camera that photographs a surrounding area of the vehicle, a three-dimensional measuring unit that measures a photographing range of the surrounding monitoring camera with the stereo camera by means for rotating the stereo camera, and generates three-dimensional information from the measurement result, a distance correction information generating unit that generates distance correction information corresponding to a photographing direction of surrounding monitoring camera from the three-dimensional information, and a distance measuring unit that measures a position of the obstacle detected with a surrounding monitoring camera.

Abstract: A medical image (24) with contrast of one imaging modality is emulated from images with other contrast characteristics. A plurality of input images (20) of a region of interest with the other contrast characteristics is received. A transform (22) which was generated by machine learning is applied to the corresponding plurality of input images (20) to generate a scalar value (38) or a vector value (38?) representing the corresponding voxel (36?) of the medical image (24) to be evaluated.

Abstract: An augmented reality-based content authoring tool is presented. A content author arranges machine-recognizable markers in a physical environment. A computing device operating as the authoring tool recognizes the markers and their arrangement based on a captured digital representation of the physical environment. Once recognized, augmented reality primitives corresponding to the markers can be bound together via their primitive interfaces to give rise to a content set. The individual primitives and content set are instantiated based on the nature of the marker's arrangement.

Abstract: A method for finding a best matching string among a set of strings for a reference string includes representing, for each of the set of strings paired with the reference string, a dynamic programming problem for calculating a final alignment score as a matrix of cells, and calculating a current optimal alignment boundary threshold. The method also includes executing, for each string of the set of strings, a calculation of a prospective final alignment score of a candidate alignment of the each of the set of strings and the reference string for each cell. Based on determining, that prospective final alignment score improves the current optimal alignment boundary threshold, the method includes calculating a final alignment score for the string of the set of strings associated with the cell. Otherwise, the method includes aborting the calculation of a candidate alignment covering the string associated with the cell.

Abstract: An approach is provided for accurate processing and registering of media content for rendering in 3D maps and other applications. The approach includes determining at least one first pixel of at least one image that geometrically corresponds to at least one second pixel of at least one rendered three-dimensional map. Further, the approach includes processing and/or facilitating a processing of (a) the at least one first pixel; (b) the at least one second pixel; (c) metadata associated with at least one of the at least one first pixel and the second pixel; or (d) a combination thereof to determine at least one confidence value, wherein the at least one confidence value is indicative of an estimated level of geometric distortion resulting from projecting the at least one first pixel onto the at least one second pixel.

Abstract: A system and method for determining an object or product represented in an image is disclosed. The system receives a first image, determines a region of interest in the first image, determines a classification score for the region of interest using a convolutional neural network that assigns the region of interest the classification score corresponding to a class, and identifies a first product in the first image based on the classification score.

Abstract: Portable communication devices transmit digital images and their location information to a central server. If a particular location is often photographed it can be designated as a hot spot. Thereafter, if a communication device is currently transmitting from within a vicinity of the hot spot, based on the location data received from the communication device, notification data can be transmitted to the communication device for notifying the user of the hot spot. The notification data can include directional information for the user to access on the communication device for enabling the user to find the hot spot.

Abstract: A system and apparatus configured to process search radar data based on random matrix theory. During the time that the radar receiver is listening for return target echoes, the raw data stream may be fed to an analog to digital converter to create a sample voltage file. This sample voltage file may be processed by a digital signal processor that computes the eigenvalues of a sample covariance matrix generated for each pulse duration interval. The ratio of the largest to smallest eigenvalue is determined and compared to the system noise eigenvalue ratio. The sensitivity for detecting targets over the present state of the art is expected to be an approximate enhancement factor of one thousand, due to the detection of coherent energy instead of a transmitted waveform. The increase of detection distance for same radar cross section is expected to be an approximate 5.6 enhancement factor.

Abstract: Systems and methods for determining athletic attributes are disclosed. Aspects of this disclosure relate to determining athletic attributes of an athlete from image data. One or more determinations may be based alterations of image data between different images, such as alterations in pixels representing objects or portions of objects. Image data may be utilized to determine whether certain thresholds are met. Various threshold levels may be applied to one or more objects represented in the image data. Landmarks/distance calibrations may be utilized from time-stamped image data to allow for precise measuring of performance (including, but not limited to: sprint or agility times, flight time for vertical jump, distance for throws). Data retrieved or derived from the image data may be used in scoring and/or ranking athletes. Such data may be used to provide training advice or regimes to the athletes or other individuals, such as coaches or trainers.

Abstract: A system capable of determining which recognition algorithms should be applied to regions of interest within digital representations is presented. A preprocessing module utilizes one or more feature identification algorithms to determine regions of interest based on feature density. The preprocessing modules leverages the feature density signature for each region to determine which of a plurality of diverse recognition modules should operate on the region of interest. A specific embodiment that focuses on structured documents is also presented. Further, the disclosed approach can be enhanced by addition of an object classifier that classifies types of objects found in the regions of interest.

Abstract: This invention provides a video processing apparatus for suppressing a decrease in search accuracy of an object of a search target based on a feature extracted from a degraded image in a video. The video processing apparatus includes an image degradation estimator that extracts an image in an image region used as an index of image degradation from a video and estimates a degree of image degradation in the image region, a feature extractor that extracts a feature of an object of a search target from the video, a feature database that stores the feature of the object of the search target, and a matcher that controls matching between the extracted feature and the stored feature based on the degree of image degradation estimated by the image degradation estimator, when searching for the object from the video.

Abstract: An augmented reality-based content authoring tool is presented. A content author arranges machine-recognizable markers in a physical environment. A computing device operating as the authoring tool recognizes the markers and their arrangement based on a captured digital representation of the physical environment. Once recognized, augmented reality primitives corresponding to the markers can be bound together via their primitive interfaces to give rise to a content set. The individual primitives and content set are instantiated based on the nature of the marker's arrangement.

Abstract: A method of combining 2D images into a 3D image includes providing a coordinate measurement device and a triangulation scanner having an integral camera associated therewith, the scanner being separate from the coordinate measurement device. In a first instance, the coordinate measurement device determines the position and orientation of the scanner and the integral camera captures a first 2D image. In a second instance, the scanner is moved, the coordinate measurement device determines the position and orientation of the scanner, and the integral camera captures a second 2D image. A cardinal point common to the first and second images is found and is used, together with the first and second images and the positions and orientations of the scanner in the first and second instances, to create the 3D image.

Abstract: A feature amount generation device includes a feature point detecting unit which detects a feature point from an input image, a pixel information generating unit which generates pixel information from pixel values of the surroundings of the feature point in the input image, a feature point pixel information distribution generating unit which generates a first distribution of pixel information of a first region including the feature point, a sub-feature point detecting unit which detects sub-feature points that are feature points in the surroundings of the feature point, a sub-feature point pixel information distribution generating unit which generates second distributions of pixel information of second regions including the sub-feature points, and a feature amount generating unit which generates a feature amount at the feature point based on the first distribution and the second distributions.

Abstract: Interaction-based ecosystems are presented. Interaction analysis engine analyze media content to derive a set of media features. The engine can then identify one or more interaction objects (e.g., transactions, searches, game play, etc.) based on the set of media features. Relevant interaction objects can then be instantiated as persistent available or active points of interaction readily accessed by a consumer. The consumer need only capture a digital representation of the content via a user device, a smart phone for example. A second set of media features can be derived from the digital representation and the second set of media features can then be used to find the instantiated interactions.

Abstract: Provided are a system and method for image sharpening is provided that involves capturing an image, and then decomposing the image into a plurality of image-representation components, such as RGB components for example. Each image-representation component is transformed to obtain an unsharpened multi-resolution representation for each image-representation component. A multi-resolution representation includes a plurality of transformation level representations. Sharpness information is transported from an unsharpened transformation level representation of a first one of the image-representation components to a transformation level representation of an unsharpened multi-resolution representation of a second one of the image-representation components to create a sharpened multi-resolution representation of the second one of the image-representation components. The sharpened multi-resolution representation of the second one of the image-representation components is then transformed to obtain a sharpened image.

Abstract: A system for marking video comprises an input interface, a processor, and an output interface. The input interface is configured to receive event recorder data from an event data recorder and to receive an indication of an event of interest. The processor is configured to determine a start mark and an end mark for the event of interest. The output interface configured to provide the start mark and the end mark to a continuous video recorder of the vehicle.

Abstract: SIFT features are selected from an input image. A SIFT procedure is applied to the input image to obtain candidate keypoints. For each candidate keypoint, there is calculation of a first Laplacian value (Lu) for pixels in an upper Scale Space and a second Laplacian value (Ll) for pixels in a lower Scale Space, based on the position of the candidate keypoint. A candidate keypoint is discarded if a Laplacian value Lc of the keypoint position is less than or equal to either of Lu or Ll. In the case that the candidate keypoint is not discarded, the candidate keypoint's Laplacian strength (Ls) is calculated, based on a relative change in Laplacian value from Lc to Lu and from Lc to Ll. One or more candidate keypoints are selected as SIFT features based on the corresponding Laplacian strength.