Abstract: This invention relates to an artificial memory system and a method of continuous learning for predicting and anticipating human operator's action as response to ego-intention as well as environmental influences during machine operation. More specifically the invention relates to an architecture with artificial memory for interacting with dynamic behaviors of a tool and an operator, wherein the architecture is a first neural network having structures and mechanisms for abstraction, generalization and learning, the network implementation comprising an artificial hierarchical memory system.

Abstract: An optical rangefinder based on time-of-flight measurement, radiates pulsed light toward an object (70), and receives reflected light from the object, the receiver operating in a photon counting mode, so as to generate a pulse for a detected photon. There is a variable probability of a photon detection on the receiver, and a controller (370, 380, 390; 365, 470, 475, 380, 390; 570, 580, 590, 390) controls the photon detection probability of the receiver, based on a light level. By controlling the detection probability according to a light level, the receiver can have an increased dynamic range, and without the expense of using optical components. This can apply even while detecting very weak signals since the receiver can still be in a photon counting mode while the detection probability is controlled. The light level can be indicated by an output of the receiver itself, or by another detector external to the receiver.

Abstract: This invention relates to an artificial memory system and a method of continuous learning for predicting and anticipating human operator's action as response to ego-intention as well as environmental influences during machine operation. More specifically the invention relates to an architecture with artificial memory for interacting with dynamic behaviors of a tool and an operator, wherein the architecture is a first neural network having structures and mechanisms for abstraction, generalization and learning, the network implementation comprising an artificial hierarchical memory system.

Abstract: A method (100) and system (300) is described for processing video data comprising a plurality of images. The method (100) comprising obtaining (104, 106), for each of the plurality of images, a segmentation in a plurality of regions and a set of keypoints, and tracking (108) at least one region between a first image and a subsequent image resulting in a matched region in the subsequent image taking into account a matching between keypoints in the first image and the subsequent image. The latter results in accurate tracking of regions. Furthermore the method may optionally also perform label propagation taking into account keypoint tracking.

Abstract: A submillimeter wavelength radar system has a receiver (20, 27, 90) for receiving and downconverting signals from content in a field of view of the system and a signal processor (30) arranged to determine information about the content from the downconverted signals, the radar system being arranged to obtain signals of the same points in the field of view from different illumination or receiving angles by having multiple illumination or receive positions, and the signal processor being arranged to use the determined information from the signals from the two or more angles to determine location or orientation of the content. By using information from different angles, it becomes possible to address or overcome the drawback of submillimeter wavelengths that most of the reflection is specular and so only surfaces of an object facing the radar system are detectable, meaning that many objects are unrecognisable.

Abstract: A signal processor (30) for a submillimeter wavelength active radar system (10, 20, 30) processes signals received and downconverted by the radar system, the downconverted signals corresponding to a given pixel of the field of view having time varying amplitude and phase components which have a periodic component which is dependent on content. Information about the content is discriminated from the periodic component. By using phase rather than only amplitude, there is additional information in the downconverted signals. The phase is more sensitive to changes in the content such as objects, background and atmospheric conditions, than amplitude alone. The phase information enables the periodic component to be retained which can be characteristic of the content owing to content flutter, changes in submillimeter standing waves and interference fringes in received reflections of submillimeter illumination if surface layers have a thickness of a number of half wavelengths.

Abstract: A method for operating on images is described for interest point detection and/or description working under different scales and with different rotations, e.g. for scale-invariant and rotation-invariant interest point detection and/or description.

Abstract: An antenna (80,90) has a one dimensional or multidimensional array of elements (20,40), wherein spacings between successive elements of at least part of the array are non periodic and correspond to a series of multiples of a unit spacing, the multiples following a Fibonacci sequence. Two dimensional arrays can be arranged as a Fibonacci grid or as a Fibonacci square tiling. The number of elements can be reduced for a given measure of resolution, while still enabling the signal being transmitted or received to have a peak in a single unique direction and thus form a beam. Furthermore, since there will be some elements clustered close together and a few which are well spaced, it can be more suitable for vehicles (30) than a regularly spaced array. It can be used as a transmit antenna or as a receive antenna for a submillimeter radar system.

Abstract: We describe methods of characterizing a set of images to determine their respective illumination, for example for recovering the 3D shape of an illuminated object. The method comprises: inputting a first set of images of the object captured from different positions; determining frontier point data from the images, this defining a plurality of frontier points on the object and for each said frontier point a direction of a normal to the surface of the object at the frontier point, and determining data defining the image capture positions; inputting a second set of images of said object, having substantially the same viewpoint and different illumination conditions; and characterizing the second set of images said frontier point data to determine data comprising object reflectance parameter data (?) and, for each image of said second set, illumination data (L) comprising data defining an illumination direction and illumination intensity for the image.

Abstract: A method (400) of processing signal outputs of a plurality of topologically distinct sensors in response to stimuli is described. The method comprises obtaining (402) a plurality of temporal sensor outputs in parallel. Thereafter, features are extracted (406), the features having dynamic behavior pattern. The extraction is performed in a topology consistent way by arithmetic processing in parallel of neighboring temporal sensor outputs. Furthermore, a quality of the extracted features is being determined.

Abstract: A method for operating on images is described for interest point detection and/or description working under different scales and with different rotations, e.g. for scale-invariant and rotation-invariant interest point detection and/or description.

Abstract: The invention relates to a vehicle or traffic control method and to a vehicle or traffic control system. The vehicle or traffic control method comprises the steps: a) estimating actual and/or future behavior of a first traffic participant and of a second traffic participant, respectively, the second traffic participant being different from the first traffic participant, b) estimating a trajectory to be taken by the first traffic participant and/or a trajectory to be taken by the second traffic participant, c) determining risk of collision of the first traffic participant relative to the second traffic participant by calculating information adapted for risk assessment of collision of the first traffic participant relative to the second traffic participant, and d) controlling the behavior of the first traffic participant based on the information provided after step a), step b) and/or step c).

Abstract: Methods and apparatus for operating on images are described, in particular methods and apparatus for interest point detection and/or description working under different scales and with different rotations, e.g. for scale-invariant and rotation-invariant interest point detection and/or description. The present invention can provide improved or alternative apparatus and methods for matching interest points either in the same image or in a different image. The present invention can provide alternative or improved software for implementing any of the methods of the invention. The present invention can provide alternative or improved data structures created by multiple filtering operations to generate a plurality of filtered images as well as data structures for storing the filtered images themselves, e.g. as stored in memory or transmitted through a network. The present invention can provide alternative or improved data structures including descriptors of interest points in images, e.g.

Abstract: An optical rangefinder based on time-of-flight measurement, radiates pulsed light toward an object (70), and receives reflected light from the object, the receiver operating in a photon counting mode, so as to generate a pulse for a detected photon. There is a variable probability of a photon detection on the receiver, and a controller (370, 380, 390; 365, 470, 475, 380, 390; 570, 580, 590, 390) controls the photon detection probability of the receiver, based on a light level. By controlling the detection probability according to a light level, the receiver can have an increased dynamic range, and without the expense of using optical components. This can apply even while detecting very weak signals since the receiver can still be in a photon counting mode while the detection probability is controlled. The light level can be indicated by an output of the receiver itself, or by another detector external to the receiver.

Abstract: Method for processing a sequence of at least two image datasets each having at least two pixels, each pixel having an intensity value by calculating a structure tensor for each pixel in an image dataset included in the sequence of image datasets; calculating values in a summation kernel based on the structure tensor for each pixel in the image dataset; calculating a weighted intensity value for each pixel in the first image dataset, using as weights the values in the summation kernel; storing the weighted intensity value for each pixel in the image dataset as a processed intensity value for each corresponding pixel in a processed output image dataset; rotating a local coordinate system in which the summation kernel is described so that the coordinate axes of said local coordinate system coincide with the directions of the eigenvectors of the structure tensor.

Abstract: A signal processor (30) for a submillimeter wavelength active radar system (10, 20, 30) processes signals received and downconverted by the radar system, the downconverted signals corresponding to a given pixel of the field of view having time varying amplitude and phase components which have a periodic component which is dependent on content. Information about the content is discriminated from the periodic component. By using phase rather than only amplitude, there is additional information in the downconverted signals. The phase is more sensitive to changes in the content such as objects, background and atmospheric conditions, than amplitude alone. The phase information enables the periodic component to be retained which can be characteristic of the content owing to content flutter, changes in submillimeter standing waves and interference fringes in received reflections of submillimeter illumination if surface layers have a thickness of a number of half wavelengths.

Abstract: An antenna (80,90) has a one dimensional or multidimensional array of elements (20,40), wherein spacings between successive elements of at least part of the array are non periodic and correspond to a series of multiples of a unit spacing, the multiples following a Fibonacci sequence. Two dimensional arrays can be arranged as a Fibonacci grid or as a Fibonacci square tiling. The number of elements can be reduced for a given measure of resolution, while still enabling the signal being transmitted or received to have a peak in a single unique direction and thus form a beam. Furthermore, since there will be some elements clustered close together and a few which are well spaced, it can be more suitable for vehicles (30) than a regularly spaced array. It can be used as a transmit antenna or as a receive antenna for a submillimeter radar system.

Abstract: A submillimeter wavelength radar system has a receiver (20, 27, 90) for receiving and downconverting signals from content in a field of view of the system and a signal processor (30) arranged to determine information about the content from the downconverted signals, the radar system being arranged to obtain signals of the same points in the field of view from different illumination or receiving angles by having multiple illumination or receive positions, and the signal processor being arranged to use the determined information from the signals from the two or more angles to determine location or orientation of the content. By using information from different angles, it becomes possible to address or overcome the drawback of submillimeter wavelengths that most of the reflection is specular and so only surfaces of an object facing the radar system are detectable, meaning that many objects are unrecognisable.

Abstract: A method is disclosed for enhancing the quality of an image dataset, such as, e.g., reducing the noise in a noisy image data set and increasing the contrast in the image data set. The method may be used for processing a sequence of image datasets, e.g. night vision image datasets, wherein said sequence comprises at least two image datasets each having at least two pixels, and wherein each pixel has an intensity value.

Abstract: A method (100) and system (300) is described for processing video data comprising a plurality of images. The method (100) comprising obtaining (104, 106), for each of the plurality of images, a segmentation in a plurality of regions and a set of keypoints, and tracking (108) at least one region between a first image and a subsequent image resulting in a matched region in the subsequent image taking into account a matching between keypoints in the first image and the subsequent image. The latter results in accurate tracking of regions. Furthermore the method may optionally also perform label propagation taking into account keypoint tracking.