Image analysis – Image compression or coding – Adaptive coding
Reexamination Certificate
1999-02-24
2002-03-12
Johnson, Timothy M. (Department: 2623)
Image analysis
Image compression or coding
Adaptive coding
C375S240080, C375S240210, C375S240290, C382S175000, C382S243000, C382S311000
Reexamination Certificate
active
06356664
ABSTRACT:
CROSS-REFERENCES TO RELATED APPLICATIONS
U.S. Pat. No. 5,793,895—granted Aug. 11, 1998 to Yuan-Chi Chang et al, entitled INTELLIGENT ERROR RESILIENT VIDEO ENCODER, and assigned to the International Business Machines Corporation, discloses a video compression method that differs from but could be applied together with the data reduction technique disclosed in this application.
GLOSSARY OF TERMS
I-Frame: stand-alone image data, usually compressed using JPEG compression techniques, containing all information needed to fully recreate the associated image
P-Frame: image data representing a “forward predicted” video frame, that is coded using motion vectors and error blocks to allow reconstruction from a reference frame, actually a previous P or I type frame in the same stream of image frame data
B-Frame: image data coded like P-frames, but with a bi-directional context of referral wherein motion vectors can refer to either a previous frame, a subsequent frame, or both.
MPEG: A video compression technique that maps signals representing a series of standard motion picture frames into a series of I, B, and P frames.
MJPEG: A video compression technique in which individual frames are separately compressed using the MPEG standard
NTSC: National Television Standards Committee. Also used as a name for that committee's standard for broadcast quality video capture and reproduction.
PAL: Phase Alternating Line system; a European standard for broadcast quality video capture, broadcast and reproduction.
MB/sec. Megabits per second; a unit of transmission bandwidth
FIELD OF THE INVENTION
The invention pertains to a weighting technique for reducing video data to facilitate efficient handling thereof either in transmission over communication media or in facilities storing the data.
BACKGROUND OF THE INVENTION
For many years, video information has been recorded in analog form. More recently, digital cameras have been recording such information in digital forms. In many instances, the information is distributed over digital communication networks to receiving sites containing computer monitors and television sets. Such transmissions—which are used, for example, in TV broadcasts, video conferences, and video transfers over the Internet—usually require sizable amounts of storage and transmission bandwidth; since a typical file transmitted in this manner may contain hundreds or thousands of megabytes, and require real time transport speeds on the order of 15 megabits per second to be synchronously displayable at receiving destinations.
Today, common standards for video transmission are represented by commercial television standards such as NTSC and PAL. Digital forms of these standards, such as D
1
, require significant transmission bandwidth (e.g. D
1
needs on the order of 270 MB/sec.).
Accordingly, there is significant commercial interest in being able to reduce storage and transmission bandwidth requirements associated with such handling of video data. Some of these requirements are eased by use of existing data compression encoding standards such as MPEG and MJPEG, and variants thereof. Such usage generally involves compression of all bits in each video frame based solely on changes from prior frames; and it allows for moderately efficient compression of television quality video. More aggressive compression technologies, such as that disclosed in the above cross-referenced US Patent to Chang et al, allow for even greater reductions in bandwidth usage with loss of picture size and quality.
For example, a video of a celebrity addressing a large crowd of people in a public square might include large amounts of vehicular traffic and other motion. MPEG would use a considerable portion of its allotted bandwidth to capture such traffic and motion in its transmission of the video in compressed form. MJPEG would use more bandwidth than MPEG to transmit a given video series with comparable quality. The process described in the cross-referenced patent to Chang et al uses a version of MPEG with special handling effects to achieve additional compression.
Existing video compression standards, such as MPEG and MJPEG, tend to assign equal importance to all bits or pixels of information in a video frame, and therefore to apply identical compression treatments to all such bits or pixels. Other known techniques for reducing video data allow for apportioning available bandwidth by applying different intensities of compression to foreground and background portions of each transmitted frame; with the effect of producing frame images at receiving sites that show details of moving objects in the foreground with high clarity and details of moving objects in the background with lesser clarity.
We, on the other hand, use a methodology involving selective reductions in rates of transfer of selected picture elements in video frames, based on an importance weighting scheme that we believe is novel. Using this scheme, portions of a frame are assigned different importance weights, and the weighted portions are selectively sampled and transmitted at rates proportional to their assigned weights. In effect, this varies the transmission bandwidth allocated to such portions; so that portions weighted most important are allocated more bandwidth than portions with lesser weights. The portions having lesser weights are relatively compressed in relation to the portions of higher weight, but such compression is ancillary to the main objective of having the transmitted and reproduced motion picture resulting from this technique show details of most important picture elements with greatest clarity and details of less important picture elements with less clarity (and even deliberately blurred or obscured in certain instances).
SUMMARY OF THE INVENTION
Our invention concerns a method and apparatus for reducing video data; e.g. to reduce bandwidth required for picture transmission or to reduce storage capacity needed for image storage. In this arrangement, portions of video (or motion picture) frames, taken from a common source such as a camera or a replayed video tape, are assigned a plurality of different importance ratings—ranging e.g. from most important to less important to least important. The weighted portions are then sampled for transmission (or for transfer to compacted storage) at different frame rates proportionate to their assigned importance weights. Portions assigned highest importance weight are sampled nominally at the same rate as the frame sampling frequency of the original video source, and portions assigned lesser weights are sampled at a rate less than the original frame sampling frequency.
It is important to understand that although picture portions assigned other than a least important weight generally will contain moving or changing image elements, the motion or change per se is not a criteria for the assigned importance classification. Rather, particular facets or characteristics of the elements are key to their weight assignments. Thus, with this methodology, a picture may contain first and second dynamically changing portions wherein the first portion is weighted as most important and the second as less and possibly least important (i.e. the second could be weighted at the same importance level as areas containing completely static elements).
Furthermore, in this technique, tagged portions that represent changing background parts of a picture frame could be assigned importance equal to weights assigned to changing foreground parts of the same frame. The effect of sampling the weighted portions at rates proportional to their assigned weights is to allocate transmission bandwidth to these portions proportionate to their weights; the more important, the more bandwidth. As a consequence of their reduced sampling rates, transmissions of signals representing picture portions having weights less than the highest assignable weight are effectively compressed in relation to transmissions of picture portions assigned the highest weight. However, such compression is considered incidental or ancillary to the previously me
Dunn James M.
Stern Edith H.
Willner Barry E.
International Business Machines - Corporation
Johnson Timothy M.
Lieber Robert
Tomlin Richard A.
LandOfFree
Selective reduction of video data using variable sampling... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Selective reduction of video data using variable sampling..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Selective reduction of video data using variable sampling... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2886130