Abstract: A method for calibrating a vehicle cabin camera having: a pitch, yaw and roll angle; and a field of view capturing vehicle cabin features which are symmetric about a vehicle longitudinal axis comprises: selecting points from within an image of the vehicle cabin and projecting the points onto a 3D unit circle in accordance with a camera projection model. For each of one or more rotations of a set of candidate yaw and roll rotations, the method comprises: rotating the projected points with the rotation; flipping the rotated points about a pitch axis; counter-rotating the projected points with an inverse of the rotation; and mapping the counter-rotated points back into an image plane to provide a set of transformed points. A candidate rotation which provides a best match between the set of transformed points and the locations of the selected points in the image plane is selected.

Abstract: A method for calibrating a vehicle cabin camera having: a pitch, yaw and roll angle; and a field of view capturing vehicle cabin features which are symmetric about a vehicle longitudinal axis comprises: selecting points from within an image of the vehicle cabin and projecting the points onto a 3D unit circle in accordance with a camera projection model. For each of one or more rotations of a set of candidate yaw and roll rotations, the method comprises: rotating the projected points with the rotation; flipping the rotated points about a pitch axis; counter-rotating the projected points with an inverse of the rotation; and mapping the counter-rotated points back into an image plane to provide a set of transformed points. A candidate rotation which provides a best match between the set of transformed points and the locations of the selected points in the image plane is selected.

Abstract: A method for reducing camera motion blur comprises, before acquiring an image frame for a video stream, a camera measurement unit measuring data related to a camera module motion during a time window; determining camera module motion based on the measured data and predicting a camera motion blur during acquisition of the image frame based at least on the determined camera module motion and the lens projection model; determining whether the predicted camera motion blur exceeds a threshold; in response to determining that the predicted camera motion blur exceeds the threshold, determining a reduction of the provisional exposure time determined to acquire the image frame so that the predicted camera motion blur reaches the threshold, determining whether a corresponding increase in the provisional gain determined to acquire the image frame is below a maximum gain value, adjusting the provisional exposure time and gain, and acquiring the image frame.

Abstract: Disclosed is a multi-modal convolutional neural network (CNN) for fusing image information from a frame based camera, such as, a near infra-red (NIR) camera and an event camera for analysing facial characteristics in order to produce classifications such as head pose or eye gaze. The neural network processes image frames acquired from each camera through a plurality of convolutional layers to provide a respective set of one or more intermediate images. The network fuses at least one corresponding pair of intermediate images generated from each of image frames through an array of fusing cells. Each fusing cell is connected to at least a respective element of each intermediate image and is trained to weight each element from each intermediate image to provide the fused output. The neural network further comprises at least one task network configured to generate one or more task outputs for the region of interest.

Abstract: A system for generating motion blur comprises: a frame camera, an event camera and an accumulator for accumulating event information from a plurality of events occurring within a window around the exposure time of an image frame in a plurality of event frames. A processor determines from the events in at least a first of the plurality of event frames, one or more areas of movement within the field of view of the event camera; determines from the events in at least a second of the plurality of event frames, a direction of movement for the one or more areas of movement; and applies blur in one or more areas of the image frame corresponding to the one or more determined areas of movement in accordance with at least the direction of movement for each of the one or more areas of movement to produce a blurred image.

Abstract: A decompression apparatus comprises a number of stages including: a first stage which always reads a binary symbol from a first stage indicator file for each symbol which is to be decoded; one or more mid stages which conditionally read a binary symbol from successive indicator files based on the value of the last symbol read from a previous indicator file; and a final stage which conditionally reads a symbol from a reduced file based on the value of the last symbol read from the last stage indicator file.

Abstract: A method for identifying a gesture from one of a plurality of dynamic gestures, each dynamic gesture comprising a distinct movement made by a user over a period of time within a field of view of an image acquisition device comprises iteratively: acquiring a current image from said image acquisition device at a given time; and passing at least a portion of the current image through a bidirectionally recurrent multi-layer classifier. A final layer of the multi-layer classifier comprises an output indicating a probability that a gesture from the plurality of dynamic gestures is being made by a user during the time of acquiring the image.

Abstract: A method for training a neural network for detecting a plurality of classes of object within a sample comprises providing a training data set comprising a plurality of samples, each annotated according to whether the samples include labelled objects of interest. In a first type of samples, all objects of interest are labelled according to their class and comprise a foreground of the samples, the remainder of the samples comprising background. In a second type of samples, some objects of interest are labelled in a foreground and their background may comprise unlabelled objects. A third type of samples comprise only background comprising no objects of interest. Negative mining is only performed on the results of processing the first and third types of samples.

Abstract: Techniques and arrangements that utilize speckle imaging and autocorrelation to estimate the PSF of an image sensor for a digital imaging apparatus, e.g., a camera or a scanner. In particular, a system of components described herein is a simple arrangement that does not require a complex setup. Therefore, the system is portable and easy to set up. Additionally, by utilizing autocorrelation, the calculations of PSF using data obtained by the system are simplified.

Abstract: A method of tracking an object across a stream of images comprises determining a region of interest (ROI) bounding the object in an initial frame of an image stream. A HOG map is provided for the ROI by: dividing the ROI into an array of M×N cells, each cell comprising a plurality of image pixels; and determining a HOG for each of the cells. The HOG map is stored as indicative of the features of the object. Subsequent frames are acquired from the stream of images. The frames are scanned ROI by ROI to identify a candidate ROI having a HOG map best matching the stored HOG map features. If the match meets a threshold, the stored HOG map indicative of the features of the object is updated according to the HOG map for the best matching candidate ROI.

Abstract: A vehicle occupant monitoring system, OMS, comprises an image acquisition device with a rolling shutter image sensor configured to selectively operate in either: a full-resolution image mode where an image frame corresponding to the full image sensor is provided; or a region of interest, ROI, mode, where an image frame corresponding to a limited portion of the image sensor is provided. An object detector is configured to receive a full-resolution image from the image sensor and to identify a ROI corresponding to an object of interest within the image. A controller is configured to obtain an image corresponding to the ROI from the image sensor operating in ROI mode, the image having an exposure time long enough for all rows of the ROI to be illuminated by a common pulse of light from at least one infra-red light source and short enough to limit motion blur within the image.

Abstract: A method at a first participant's client conferencing system in a videoconference comprises receiving, from a second client conferencing system, at least one first video frame of a first video signal including an image of the second participant looking at a third participant, and first metadata associated with the first video frame and including an identity of the third participant. The image of the second participant is modified in the first video frame so that the first video frame is displayed on a first area of the client conferencing system with the second participant looking at a second area of the first display configured for displaying a second video signal of the third participant identified by the first metadata.

Abstract: A decompression apparatus comprises a number of stages including: a first stage which always reads a binary symbol from a first stage indicator file for each symbol which is to be decoded; one or more mid stages which conditionally read a binary symbol from successive indicator files based on the value of the last symbol read from a previous indicator file; and a final stage which conditionally reads a symbol from a reduced file based on the value of the last symbol read from the last stage indicator file.

Abstract: An auto-exposure module determines a first set of exposure parameters to acquire a first image. The first image is acquired using the first set of exposure parameters. The module determines a target image histogram; calculates an image histogram based on intensity values of the pixels of the first image; determines an integral of the image histogram and of the target image histogram for a range of pixel intensity values; calculates a transfer curve for transforming the integral of the image histogram to match the integral of the target image histogram; calculates a slope of a line fitting at least a portion of the transfer curve; determines a correction factor based on the calculated slope; and adjusts the first set of exposure parameters according to the correction factor. A second image is then acquired using the adjusted first set of exposure parameters.

Abstract: A vehicle occupant monitoring system, OMS, comprises: an image acquisition device comprising an image sensor and a lens assembly having a varying transmissivity across a field of view of the image sensor; at least one infra-red, IR, light source disposed within a cabin of the vehicle and being configured to illuminate at least one occupant of the vehicle with varying illumination across the field of view of the image sensor; and an image processing pipeline configured to obtain and pre-process an image acquired from the image sensor in accordance with a lens shading map and a cabin illumination map in order to compensate for both the varying transmissivity and the varying illumination in order to provide a more uniformly illuminated image to a controller for further analysis.

Abstract: A vehicle occupant monitoring system, OMS, comprises an image acquisition device with a rolling shutter image sensor comprising an array of sub-pixels which are respectively selectively sensitive to: red and infra-red; blue and infra-red; and green and infra-red light. The device is configured to selectively operate in either: a colour mode where a multi-plane image frame corresponding to the full image sensor is provided, each plane derived from red, green or blue sensitive sub-pixels respectively; or a monochrome mode, where sensor information from sub-pixels is aggregated to provide a single image plane.

Abstract: A vehicle occupant monitoring system, OMS, comprises an image acquisition device with a rolling shutter image sensor configured to selectively operate in either: a full-resolution image mode where an image frame corresponding to the full image sensor is provided; or a region of interest, ROI, mode, where an image frame corresponding to a limited portion of the image sensor is provided. An object detector is configured to receive a full-resolution image from the image sensor and to identify a ROI corresponding to an object of interest within the image. A controller is configured to obtain an image corresponding to the ROI from the image sensor operating in ROI mode, the image having an exposure time long enough for all rows of the ROI to be illuminated by a common pulse of light from at least one infra-red light source and short enough to limit motion blur within the image.

Abstract: A method for reducing camera motion blur comprises, before acquiring an image frame for a video stream, a camera measurement unit measuring data related to a camera module motion during a time window; determining camera module motion based on the measured data and predicting a camera motion blur during acquisition of the image frame based at least on the determined camera module motion and the lens projection model; determining whether the predicted camera motion blur exceeds a threshold; in response to determining that the predicted camera motion blur exceeds the threshold, determining a reduction of the provisional exposure time determined to acquire the image frame so that the predicted camera motion blur reaches the threshold, determining whether a corresponding increase in the provisional gain determined to acquire the image frame is below a maximum gain value, adjusting the provisional exposure time and gain, and acquiring the image frame.

Abstract: An image processing system comprising a processer configured to receive a sequence of images frames from an image acquisition device and configured to: analyze at least a currently acquired image frame to determine if activity is occurring in an environment with a field of view of the image acquisition device; responsive to analyzing a subsequent image frame acquired after the currently acquired image frame and determining that no activity is occurring in the environment, retrieve an image frame acquired before the currently acquired image frame which has been analyzed and where it has been determined that no activity is occurring in the environment; analyze the subsequent image frame and the retrieved image frame to identify a state of one or more objects within the field of view of the image acquisition device; and responsive to a change in state of the one or more objects, notify a user accordingly.

Abstract: A method at a first participant's client conferencing system in a videoconference comprises receiving, from a second client conferencing system, at least one first video frame of a first video signal including an image of the second participant looking at a third participant, and first metadata associated with the first video frame and including an identity of the third participant. The image of the second participant is modified in the first video frame so that the first video frame is displayed on a first area of the client conferencing system with the second participant looking at a second area of the first display configured for displaying a second video signal of the third participant identified by the first metadata.