On December 27, the US Patent & Trademark Office published Apple’s patent application titled Pulldown correction for progressive display of audiovisual recordings . Apple’s patent specifically covers technology that would primarily appeal to professional video markets such as Hollywood and television broadcasters.
Partial Patent Background
When a film or other audiovisual recording is transferred from its original format to a compressed format, it is often converted from one frame rate to another. For example, a motion picture is typically recorded at 24 frames per second (fps) in progressive format, but may be converted to 30 fps for distribution on DVD format or for television broadcast, typically using interlaced displays. An original recording may also be made at other frame rates, such as home video recordings which are typically made at 30 fps in interlaced format. Prior to encoding, an original recording may also be preprocessed, for example to perform noise reduction or frame rate conversion, and edited, for example to insert scene changes.
To compensate for the disparity between the original recording’s frame rate and the rate at which it may later be displayed, various techniques of repeating and/or dropping portions of frames are used. The most common technique, used to convert from 24 fps progressive to 30 fps interlaced, is the “3-2 pulldown.” Each original progressive frame is first converted to a set of two fields. For every other group of two fields one field is repeated, resulting in a group of three fields followed by a group of two fields, i.e., a 3-2 pattern. The resulting video sequence can then be displayed at 30 fps on an interlaced display device without introducing visual artifacts. Various other conversion techniques may be used.
However, cadence detection and pulldown correction may be inaccurate due to noise in the original video sequence, which can lead to incorrect processing. In addition, an encoder may be instructed to encode every field, regardless of whether it is a repeated field. In some cases, an encoder may insert flags into the video stream to indicate when a field is repeated, allowing a decoder to avoid decoding the same field twice. Such methods may be error-prone if the encoder incorrectly identifies repeated fields or does not mark fields consistently.
When decoding a stream generated by an encoder that did not perform pulldown correction or performed pulldown correction incorrectly, it may be desirable for a decoder to reliably identify repeated fields regardless of the presence or absence of repeated-field indicators in the stream. A decoder capable of performing pulldown correction may retrieve the original progressive content, thus reducing visual artifacts resulting from improper matching of fields when the video is displayed on a progressive device. By properly identifying and dropping appropriate fields in a video stream, visual artifacts may be reduced.
Method to Detect a Cadence of a Video Sequence
Embodiments of the present invention relate to methods and systems that may detect cadences and duplicate fields in a video sequence, and correct the sequence to retrieve the original progressive content. At some point in a video distribution system, video processing equipment may process a video sequence for which there is no information to indicate whether the video sequence is represented in interlaced format or progressive format or for which information exists but is coded incorrectly. The present invention provides techniques for a video processing system to identify a cadence from the content of the video sequence itself. Based on he identified cadence, if any, the video processing system may control its own operation, for example, during video coding or video display. In some embodiments, the invention provides techniques to determine the video type, such as progressive or interlaced, of the original video sequence. If the original video sequence is progressive, the original progressive content may be recovered.
Apple’s patent FIG. 2 illustrates a method to detect a cadence of a video sequence according to an embodiment of the present invention.
Other Embodiments of the Invention
Embodiments of the present invention may compare consecutive fields of the same parity to identify repeated fields. A similarity measure may be calculated that provides an indication of how similar two fields of the same parity are. If the similarity measure meets a threshold, one of the fields may be identified as a repeat field. In some embodiments, an adaptive threshold may be used.
Embodiments of the present invention measure activity within display fields and across consecutive display fields to determine whether two consecutive fields can be paired as a progressive frame or not. By making this determination across multiple pairs of fields, a video cadence may be detected. The relation between the field and frame activities may indicate whether two consecutive display fields may be grouped together as a frame.
In some embodiments, the field and frame activity comparison may be correlated with the similarity measure to detect the cadence of a video stream appropriately. The system may further detect interruptions in the video cadence and group fields appropriately in response to scene cuts or other discontinuities in the video sequence.
Apple’s patent FIG. 8 shows the reconstruction of a video stream having a scene change according to an embodiment of the present invention.
Apple lists Cecile M. Foret (Mountain View, CA) as the inventor of this patent which was originally filed in June 2006. For the full details of this patent, click here.
NOTICE: MacNN presents only a brief summary of patents with associated graphic(s) for journalistic news purposes as each such patent application and/or grant is revealed by the U.S. Patent & Trade Office. Readers are cautioned that the full text of any patent applications and/or grants should be read in its entirety for further details.
Written and researched by Neo.
Leave a Reply
You must be logged in to post a comment.