Abstract: A Bayer-mask image is decoded by forming a decoded green array; calculating a slope at each pixel of the array, expressed as an angle; calculating an activity at each pixel; converting the slope angle and the activity into a complex number for each pixel, of modulus equal to the activity and argument equal to twice the slope angle; expressing said complex numbers in Cartesian coordinates to form a Cartesian slope signal and filtering the Cartesian slope signal with a linear spatial filter to derive a slope measure. Blue-green and red-green values are calculated and interpolated using a slope-adaptive interpolation filter steered by said slope measure.

Abstract: A first sequence of video fingerprints at a first image repetition rate is compared with a second sequence of video fingerprints at a second, different image repetition rate. A converted sequence of fingerprints is generated from the first or second sequence of fingerprints by forming a combination of the values of a number of neighboring fingerprints. The combination is a weighted average based on the temporal location of the respective neighboring fingerprints in the first sequence of fingerprints with respect to the fingerprint in the converted sequence. A correlation is then performed between the converted sequence of fingerprints and the other sequence of fingerprints.

Abstract: A method of managing visual content. In one instance, the method includes receiving a stream of video fingerprints, derived in a fingerprint generator by an irreversible data reduction process, from respective temporal regions within a particular visual content stream and at a fingerprint processor that is physically separate from the fingerprint generator via a communication network. The fingerprints are processed in the fingerprint processor to generate metadata which is not directly encoded in the fingerprints. Processing of the fingerprints includes windowing the stream of fingerprints with a time window, deriving frequencies of occurrence of particular fingerprint values or ranges of fingerprint values within each time window, determining statistical moments or entropy values of said frequencies of occurrence, comparing said statistical moments or entropy values with expected values for particular types of content, and generating metadata representing the type of the visual content.

Abstract: To generate a warning that a stereoscopic image sequence has been synthesised from a 2D image sequence, a video processor correlates left-eye image data and right-eye image data to identify any sustained temporal offset between the left-eye and right-eye image data. A measure of sustained correlation between a measured spatial distribution of horizontal disparity and a spatial model can also be used to generate the warning.

Abstract: To monitor audio-visual content which includes captions, caption fingerprints are derived from a length of each word in the caption, without regard to the identity of the character or characters forming the word. Audio-visual content is searched to identify a caption event having a matching fingerprint and missing captions; caption timing errors and caption discrepancies are measured.

Abstract: To control loudness during a junction between different types of broadcast content, such as a junction between program and commercial or promotional content, representative loudness values for content respectively before (P) and after (C) the junction are received from a playout automation system. A time-varying gain control is applied before and after the junction in order to smooth loudness around the junction. The audio gain is smoothly increased prior to the junction to a gain (P+C)/2P times higher than the original gain value. Then, the gain is reduced shortly before the junction to a value (P+C)/2C times lower than the original gain value. After the junction, the gain is returned smoothly to the original value.

Abstract: In video motion estimation an initial candidate motion vector is generated for each block and a vector error is determined as for example a DFD. Spatial gradients of pixel values are calculated and used to refine the initial candidate motion vector. The relative contribution of the spatial gradients to the refinement process depends on the vector error.

Abstract: Points of interest are identified in an image to characterize that image by dividing the image tiles consisting of adjacent pixels; finding within each tile the position of a pixel with an extremum value and ascribing that extremal value to the tile; identifying a tile with an extremal value which is more extreme than that of all adjacent tiles; and selecting as a point of interest the position within the image of the pixel with the extremum value in that identified tile.

Abstract: A watermark encoder receives a current video image together with current metadata associated with the current image. A metadata delay also makes available to the watermark decoder delayed metadata associated respectively with four or more of the preceding. Then the watermark encoder modifies pixel values of the current image not only to encode the current metadata but also the delayed metadata. At a decoder, if metadata for the current image is corrupted or missing, it can be recovered from one of the succeeding images.

Abstract: The respective occupancies of a set of audio buffers are controlled via a common target occupancy value common to all the buffers in the set. This common target may take non-integral values. For each buffer, the difference is taken between the occupancy value of the buffer and the common target occupancy value. A sample is dropped or repeated as the difference exceeds half a sample. A recursive sum is formed of the fractional parts of the difference measures from the set of buffers and the recursive sum is used to adjust the common target occupancy value, within a selected range of values.

Abstract: A method of identifying an item of video content involves providing a spatial hash value and a temporal hash value for each image in a video collection. Each hash value is based on a measure of the entropy in differences between pixel values. A table of the pair of hash values against timecode is created and ordered according to one of the hash values. A search for a given pair of hash values can then be confined to that part of the table that matches the first value.

Abstract: In the segmentation of an image into spatial regions, an image region is identified by forming difference-image pixel values from the chosen image and at least one other image in the image sequence and accumulating those difference-image pixel values horizontally and vertically. Co-located horizontal and vertical accumulations are combined and utilized in identifying the region. Difference-image pixel values can also be accumulated along other lines which pass through a point of interest.