Abstract: A method and device are provided for stitching images and blending pixel data from at least three image sensors while an object moves through their combined field of view. The combined view includes a first region where the views of at least three sensors overlap and through which the object’s track passes. The method includes obtaining time-synchronized image sequences from each sensor, detecting the object in the combined view, determining its location, and predicting its track. Based on the predicted track, a minimal set of sensor views that fully covers the track is identified. Images captured at the same time from sensors in this minimal view set are stitched and blended to form a combined image. In most of the portion of the combined image corresponding to the overlapping region, pixel blending uses data from no more than two sensors at any location.

Abstract: A computer-implemented method for enabling video recreation, comprising: acquiring, by a camera control unit, a video stream comprising a set of image data packages each comprising coded image frames; transmitting, by the control unit, at least a portion of the video stream to a client across a communication network, receiving, by the control unit, an acknowledgement message for each image data package that was successfully received by the client, storing, in a digital memory accessible to the control unit, the image data packages for which an acknowledgement message is received, and indications of successful receipt at the client of the corresponding image data packages.

Abstract: Methods, software, devices and systems encode an image frame in a video stream, such that a decoded version of the encoded image frame is provided with a first masked region. A block-aligned mask is applied to the image frame being encoded, and a pre-generated encoded image frame is used to transform the block-aligned masked region during decoding. The pre-generated encoded image frame contains motion vectors but no residual data, enabling efficient transformation of a block-aligned masked region into the first masked region with an edge that cuts across coding blocks.

Abstract: A tracked state of a tracked object in a sequence of images is determined. A region of interest (ROI) is defined in the current image. Based on at least one previous image in the sequence, motion information from the current image is aggregated to estimate the current motion of the ROI. A predicted current state of the ROI is predicted, based on a previous state of the ROI, a plurality of object tracking models, and the estimated current motion of the ROI. The predicting comprises selecting at least one of the plurality of object tracking models to be used for predicting the predicted current state. The predicted current state of the ROI is associated with the observed current state of the ROI to determine the tracked state of the tracked object.

Abstract: A method, system and software for searching for an object in a forensic search application comprises: determining a plurality of static areas in the scene; obtaining a first image depicting the scene comprising an object; determining a plurality of candidate color transforms; determining a plurality of candidate color values of the object by applying each of the plurality of candidate color transform to third pixel data from the first image, said third pixel data depicting the object in the first image; searching for an object in a forensic search application using a search request comprising a first color value; determining that the first color value matches one or more candidate color values; and returning a search response based at least in part on the object.

Abstract: Signing of an encoded video stream using a first device and second device, wherein the first device provides the original signatures of a first encoded video stream, wherein the second device applies video processing to the first encoded video stream to modify the image frames thereof and include the modified image frames into a second encoded video stream. The second device determines difference data corresponding to the video processing and includes the difference data in the second encoded video stream. On a decoder side, the decoder can authenticate the received second encoded video stream by validating the original signatures using the difference data.

Abstract: A computer-implemented method of controlling a computing hardware accelerator in a movable camera comprises: acquiring images from the movable camera; obtaining an indication of a physical state of the movable camera, indicating whether the movable camera is in a moving state or in a still state; and allocating exclusive access to the computing hardware accelerator based on the indication of the physical state. The disclosure further relates to a camera system for carrying out the computer-implemented method of controlling a computing hardware accelerator in a movable camera.

Abstract: A method, a device and a non-transitory computer-readable storage medium for setting a confidence threshold for objects detected in a region of a captured scene based on historical confidence scores for objects detected in that region.

Abstract: An arrangement for attachment of devices to a surface, such as a wall, comprises at least one device, and for each device, at least two mounting brackets. Each device has a rear side provided with a recess for each mounting bracket, wherein each mounting bracket has a male connector end and a female connector end, the male connector end being configured to be received by the female connector end of an adjoining mounting bracket associated to a neighbouring device. Each male connector end is provided with a through hole for receiving an attachment means, such as a screw. Each mounting bracket is insertable into its associated recess with an orientation in which the male connector end extends beyond a periphery of the device, and each mounting bracket is insertable into its associated recess with an orientation in which the female connector end is accessible from the periphery of the device.

Abstract: A method of providing anonymized data for facilitating reidentification in a visual tracking system, the method comprising: detecting, in images obtained from a plurality of image sources, subareas which each contain a tracking target; computing, for each subarea, a feature vector which represents a visual appearance of the tracking target therein; for a first/second subgroup of the image sources, providing first/second reidentification data items by anonymizing each feature vector using a predefined one-way function modified by a first/second tracking-rights token, and disclosing the first/second reidentification data items annotated with locations of the respective subareas to a first/second tracking client, wherein the second tracking-rights token is distinct from the first tracking-rights token; and preventing access to the feature vectors.

Abstract: A computer-implemented method in a processor device of a camera, the method comprising: acquiring image frames comprising video data, communicating with a receiver device for continuously transmitting a video stream to the receiver device over a communication network; detecting at least one object in the image frames of the video data, the detected at least one object belong to at least one predetermined object class selected as surveillance target; cropping sub-areas in the image frames of the video data, the sub-areas including the at least one detected object, and adding the cropped sub-areas to image frames of the video stream being continuously transmitted as a single video stream to the receiver device.

Abstract: A computer-implemented method for tampering detection of a microphone device, the microphone device comprising a microphone arranged in a cavity open to surroundings of the microphone device via a microphone hole, the method comprising: receiving an audio signal from the microphone; sampling the audio signal for a low frequency band; detecting an event comprising an exponentially decreasing absolute amplitude of the sampled audio signal during a time period of at least one second; and when detecting the event, outputting a tampering detection signal.

Abstract: A method for digital image stabilization (DIS) on a sequence of video frames captured by an image capture device comprises: obtaining motion values sampled from a motion signal and position values sampled from a position signal while capturing the frames, each frame being associated with a respective motion value and position value; computing, for each frame, a residual motion value to form a sequence of residual motion values; and performing DIS by (i) detecting an image feature in a reference frame, (ii) tracking the feature across subsequent frames, (iii) determining frame motion data from the feature's displacement in the subsequent frames, and (iv) generating a stabilized frame sequence based on the frame motion data. The size of a tracking window used to track the feature in each frame is set based on the residual motion values, thereby adapting the tracking to expected inter-frame motion to improve robustness and stabilization quality.

Abstract: In some implementations, a device may encode a series of images of video data into an encoded video stream including key frames and delta frames. The delta frames may be arranged according to a hierarchical prediction pattern with multiple temporal layers and a predetermined structure. The device may obtain an instruction to encode a delta frame as a skip frame. The device may determine that the skip frame is to be encoded in a first temporal layer, of the multiple temporal layers, according to the predetermined structure. The device may evaluate whether an immediately preceding frame to the skip frame is encoded, or to be encoded, in a second temporal layer, of the multiple temporal layers, that is finer than the first temporal layer. The device may perform an action based on the evaluation.

Abstract: There is provided a method, device, and system for tracking an object in a sequence of video frames by using a tracking filter which is operable in two modes. In the first mode it estimates a state vector describing a position and velocity of the object in the video frames. In the second mode it estimates a state vector describing position, velocity, and acceleration of the object in the video frames. The method includes switching from the first mode to the second mode. The switching is performed in response to a velocity difference between a velocity of the object estimated by the tracking filter and a velocity of the object indicated by motion vectors from a video encoder and being larger than a threshold.