Abstract: A method of signing prediction-coded video data comprises: obtaining a coded video sequence comprising a set of frames including at least one intra frame having independently decodable image data, and at least one inter frame, containing image data decodable by reference to another frame; generating a group hash of the coded video sequence by hashing the image data of the coded video sequence; generating a plurality of individual hashes of frame subsets including at least one frame by hashing the image data derived from each of the frame subsets, reducing the length of each of the individual hashes to a length shorter than a total length of the group hash, generating a digital signature by encrypting at least the group hash and the individual hashes with reduced lengths; providing the group hash and the reduced length individual hashes together with the digital signature and the coded video sequence.

Abstract: A method for providing a signed video bitstream suitable for transcoding from a first video format into a second video format, the method comprising: obtaining first video data in a lossy first video format; reconstructing a video sequence from the first video data; encoding the reconstructed video sequence as second video data in a second video format; computing first fingerprints of the first video data and second fingerprints of the second video data; deriving a first bitstring from the first fingerprints and a second bitstring from the second fingerprints; and providing a signed video bitstream, which includes the first video data and signature units, each signature unit including a first digital signature of the derived first bitstring and second digital signature of the derived second bitstring.

Abstract: There is provided techniques for providing digital signatures to a video stream encoded in layers. Each layer comprises encoded image frames. A digital signature for layer k=0 is provided by providing by generating signing data for layer k=0 and generating the digital signature for layer k=0. A respective digital signature for each layer k=1, . . . , K?1 is provided by, for each layer k=1, . . . , K?1 generating signing data, and generating the digital signature for layer k by encrypting the signing data, or a hash thereof, for layer k with the private key of the private-public key pair. The digital signatures for all the layers are provide to the video stream.

Abstract: A signed media bitstream comprises data units and signature units. Each signature unit is associated with one or more nearby data units and include at least one fingerprint derived from the associated data units and a digital signature of the at least one fingerprint. A storing method comprises: receiving a segment of the media bitstream; identifying N?2 instances of a repeating data unit in the received segment; pruning up to N?1 of the identified instances of the repeating data unit; and storing the received segment after pruning. A validation method comprises: receiving a segment of the media bitstream stored in accordance with the storing method; and validating a signature unit using a digital signature contained therein. Despite the pruning of the repeating data unit, the received associated data units can be successfully validated, either directly or indirectly, by means of different embodiments herein.

Abstract: A method for verifying a digital signature in a video data segment generated by a video delivery chain comprises: obtaining the video data segment comprising encoded image frames and a plurality of node device datasets associated with respective labels; wherein one or more node device datasets include respective digital signatures; wherein the one or more node device datasets comprises a first node device dataset which includes a first digital signature generated by a first node device based on a first number of node device datasets that have been last added to the video data segment before the first node device dataset; locating the labels; identifying , using the labels, the first number of node device datasets; and verifying the first digital signature. The application also discloses a method of adding a digital signature to a video data segment.

Abstract: A device and a method of signing an encoded video sequence comprising: obtaining an encoded video sequence composed of encoded image frames; generating a set of one of more frame fingerprints for each encoded image frame; generating a document comprising a header of a supplemental information unit, and a representation of the generated sets of one or more frame fingerprints; generating a document signature by digitally signing the document; generating the supplemental information unit to only consist of the document, the document signature and an indication of an end of the supplemental information unit; and signing the encoded video sequence by associating the generated supplemental information unit with the encoded video sequence.

Abstract: A method of signing video data, comprising: obtaining video data representing a video sequence; obtaining a bitstring not extracted from the video data; generating a salt by hashing the bitstring, preferably using a secret hash function; generating a first/second fingerprint either by hashing a combination of the salt and a first/second portion of the video data, or by hashing a combination of the salt and a hash of a first/second portion of the video data; and providing a signature of the video data, which includes the first fingerprint. The first and second portions may encode consecutive time segments of the video sequence, such as consecutive frames.

Abstract: A method of signing prediction-coded video data, comprising: obtaining a coded video sequence including at least one I-frame (I), which contains independently decodable image data, and at least one predicted frame (P1, P2, P3, P4), which contains image data decodable by reference to at least one other frame; generating a fingerprint (HI) of each I-frame; generating a fingerprint (HP) of each predicted frame by hashing a combination of data derived from the predicted frame and data derived from an I-frame to which the predicted frame refers directly or indirectly, wherein the fingerprint of the predicted frame is independent of any further predicted frame to which the predicted frame refers directly or indirectly; and providing a signature of the video sequence including the generated fingerprints.

Abstract: The present application relates to detecting if video images captured by a camera are depicting a live scene or a recorded video played on a monitor, display or computer screen, which is setup to hide the scene from the camera. Metadata regarding the mapping operation used to transform image data between different intensity ranges, or bit depths, is included with the video and evaluated in order to determine if a video replay attack has taken place.

Abstract: A computer implemented method for controlling a transmission of a video stream is provided. The method comprises estimating a number of bits for a group of pictures, GOP, of the video stream to be transmitted, setting a latency requirement for the transmission of the video stream, determining an average minimum video bitrate across the GOP based on the estimated number of bits and a time corresponding to a time period represented by a duration of the GOP, for video frames in the GOP setting an output bitrate for transmission of a video frame based on the latency requirement and the average minimum video bitrate, and transmitting the video frame using the output bitrate.

Abstract: One or more delay estimators are used to detect failure in a signal processing component, such as an acoustic echo canceller. A first delay estimator is used to generate i) an estimated post-processing delay between when a known audio signal was conveyed to a loudspeaker, and when a portion of the processed audio signal that includes the known audio signal was output from the signal processing component, and ii) a confidence level for the estimated post-processing delay. A failure of the signal processing component may be detected in response to the estimated post-processing delay exceeding a threshold. A second delay estimator may also be used to generate an estimated pre-processing delay and a confidence level for the estimated pre-processing delay for comparison to the estimated post-processing delay and confidence level for the estimated post-processing delay in order to provide further failure detection accuracy and specificity.

Abstract: One or more delay estimators are used to detect failure in a signal processing component, such as an acoustic echo canceller. A first delay estimator is used to generate i) an estimated post-processing delay between when a known audio signal was conveyed to a loudspeaker, and when a portion of the processed audio signal that includes the known audio signal was output from the signal processing component, and ii) a confidence level for the estimated post-processing delay. A failure of the signal processing component may be detected in response to the estimated post-processing delay exceeding a threshold. A second delay estimator may also be used to generate an estimated pre-processing delay and a confidence level for the estimated pre-processing delay for comparison to the estimated post-processing delay and confidence level for the estimated post-processing delay in order to provide further failure detection accuracy and specificity.

Abstract: An input signal is processed through noise suppression (NS) and echo control (EC) via a multipath model that reduces noise pumping effects while maintaining EC performance. A copy of a “noisy” input signal is sent to an EC component before the noisy signal is sent to a NS component, which processes the signal first, when there is a consistent noise level for estimation. The copy of the pre-processing noisy signal is sent to the EC component along with a “clean” or “noise-suppressed” signal output from the NS component. The EC component analyzes the noisy signal as if the EC was the first component in the signal chain to determine what actions to take. The EC component then applies these actions to the clean signal received from the NS component.

Abstract: An input signal is processed through noise suppression (NS) and echo control (EC) via a multipath model that reduces noise pumping effects while maintaining EC performance. A copy of a “noisy” input signal is sent to an EC component before the noisy signal is sent to a NS component, which processes the signal first, when there is a consistent noise level for estimation. The copy of the pre-processing noisy signal is sent to the EC component along with a “clean” or “noise-suppressed” signal output from the NS component. The EC component analyzes the noisy signal as if the EC was the first component in the signal chain to determine what actions to take. The EC component then applies these actions to the clean signal received from the NS component.

Abstract: A method and apparatus for finding an estimate of the delay of a signal travelling between two points. A quantity is evaluated from the signal at a final number of time instants, at both a reference point and a reception point. The values are quantized by comparison with a threshold adapted to a typical magnitude of the quantity. If the quantized values from the reception point are shifted back by the true delay with respect to the quantized values from the reference point, then certain co-occurrences of quantized values have very low probability. Hence, the best delay estimate is that shift which yields the least number of low-probability co-occurrences.

Abstract: A method and apparatus for finding an estimate of the delay of a signal travelling between two points. A quantity is evaluated from the signal at a final number of time instants, at both a reference point and a reception point. The values are quantized by comparison with a threshold adapted to a typical magnitude of the quantity. If the quantized values from the reception point are shifted back by the true delay with respect to the quantized values from the reference point, then certain co-occurrences of quantized values have very low probability. Hence, the best delay estimate is that shift which yields the least number of low-probability co-occurrences.

Abstract: The invention relates to a method, device and computer-program product for suppression of undesired temporal variations, notably flicker, in a sequence of video frames. Histogram-based and similar approaches generally do not remove all flicker. Features that are resolved only in portions of the flicker cycle will manifest themselves as residual flicker. This effect is near-universal in bright regions of a scene. The inventive solution is a mapping that aims to resolve in the output only those features that are resolved in all frames of the flicker cycle. Use of time-maximal quantile values may preserve non-resolution of such image features that are unresolved due to intermittent bright saturation. Thus, in one embodiment, a reduction of resolution is attained by means of a pixel-value mapping based on selecting, over a time window, maximal and minimal quantile values, with maximal values being used for bright spatial regions and minimal values for dark spatial regions.

Abstract: Processing multi-channel audio streams using one or more arrangements of single-channel components. Components that only process the near-end, or capture stream, such as noise suppression (NS) components, are limited in how they can be suitably arranged for processing multi-channel streams. However, components that process the near-end stream using one or more inputs from the far-end, or render stream, such as acoustic echo cancellation (AEC) and automatic gain control (AGC) components, are arranged in one or more of the ways suitable for use with multiple channels.

Abstract: An input signal is processed through noise suppression (NS) and echo control (EC) via a multipath model that reduces noise pumping effects while maintaining EC performance. A copy of a “noisy” input signal is sent to an EC component before the noisy signal is sent to a NS component, which processes the signal first, when there is a consistent noise level for estimation. The copy of the pre-processing noisy signal is sent to the EC component along with a “clean” or “noise-suppressed” signal output from the NS component. The EC component analyzes the noisy signal as if the EC was the first component in the signal chain to determine what actions to take. The EC component then applies these actions to the clean signal received from the NS component.

Abstract: Processing multi-channel audio streams using one or more arrangements of single-channel components. Components that only process the near-end, or capture stream, such as noise suppression (NS) components, are limited in how they can be suitably arranged for processing multi-channel streams. However, components that process the near-end stream using one or more inputs from the far-end, or render stream, such as acoustic echo cancellation (AEC) and automatic gain control (AGC) components, are arranged in one or more of the ways suitable for use with multiple channels.