Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
2001-04-26
2002-09-03
Lee, Richard (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C348S699000
Reexamination Certificate
active
06445741
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The invention relates to motion estimation on sub-sampled (interlaced) video data for, e.g., scanning format conversion.
The analysis of motion in a video sequence is embedded in many advanced video processing techniques, like advanced coding, noise reduction, and scan rate conversions, such as, de-interlacing and Frame Rate Conversion (FRC). In coding applications, motion information is used to minimize the prediction error and consequently the bit rate. Temporal noise filtering is very effective, but requires Motion Compensation (MC) if objects in the image move. De-interlacing, which is a prerequisite for high quality Motion Estimation (ME) and vertical scaling of interlaced video, requires information from multiple fields, which need therefore motion compensation. In case of scan rate conversion, motion information is needed for temporal interpolation that eliminates jerkiness of the motion portrayal and motion blur.
2. Description of the Related Art
Different applications may demand different types of motion estimators. In case of coding, motion estimation is based on minimizing the difference between image parts, and coding the resulting error. The resulting motion vectors do not necessarily reflect true motion within the video sequence. The motion estimation can be optimized to find the smallest error, resulting in the most efficient coding. For scan rate conversion applications, temporal interpolation is realized in the direction of the motion. Consequently, it is important that the motion vectors describe the true motion vectors within the video sequence. It is therefore not sufficient to demand the smallest difference between image parts in the motion estimation. The motion behavior in the environment is also an important factor.
Motion estimations designed for scan rate conversion can be used for coding purposes, but not immediately vice versa. The motion estimation described in this paper is therefore not restricted to scan rate conversion but designed to find the true-motion vectors.
Most video data is available in a so-called interlaced format, i.e., a format in which the odd scan lines in odd fields and even lines in even fields together constitute a frame that describes an image, but the odd and even fields are NOT describing the image at the same temporal instance. Motion in video data is preferably measured on the shortest available time interval, i.e., the field interval in case of interlaced data. Due to vertical detail in the picture (causing alias in the sub-sampled field-grid), however, it may be impossible to correctly find the vertical displacement between two consecutive fields. The alias pattern and the original detail may move differently.
Consider a stationary horizontal white line of two scanning lines width available in an interlaced video format. Odd and even fields will both show a single white line. Considering the information in two successive fields, it is ambiguous whether we deal with a one scanning line wide white line moving with a vertical velocity of 1 sample/field-period, or a stationary horizontal white line of two scanning lines width.
A common way to deal with the mentioned problem, is to first up-convert at least one of the two fields into a progressive format, and perform the motion estimation between two frames or between a frame and a field [see Ref. 2]. An alternative exists in estimating the motion applying data from three successive fields [see Ref. 3]. Finally, it is possible to hope that the low-frequency content of the image is dominant, correct the phase for those frequencies, and estimate motion more or less neglecting the problem [see Ref. 1].
A characteristic that Ref. 2 and Ref. 3 share, is that they double the access to the previous image data (from just the previous field, to the previously de-interlaced field, or the previous and the pre-previous field, respectively). Also the solution proposed in Ref. 3 introduces a new constraint in the motion estimator, in that the motion over a two-field-period has to be assumed constant. The option of Ref. 1 does not double the memory access (and capacity!) and introduces no constant motion constraints, but cannot solve the ambiguity either, as experiments on critical picture material show.
SUMMARY OF THE INVENTION
It is, inter alia, an object of the invention to reduce ambiguity in the motion estimation in a simple manner. To this end, a first aspect of the invention provides a method and a device for estimating motion vectors from sub-sampled video data. A second aspect of the invention provides an image signal display apparatus comprising such a motion vector estimating device, a motion compensated interpolator and a display device.
In a method of estimating motion vectors from sub-sampled video data in accordance with a primary aspect of the invention, first vectors are estimated between an image with a first sub-sampling phase and an earlier image sub-sampled with a second sub-sampling phase, second vectors are estimated between an image with a second sub-sampling phase and an earlier image sub-sampled with a first sub-sampling phase, and the first and second vectors are combined to obtain output motion vectors.
These and other aspects of the invention will be apparent from and elucidated with reference to the embodiments described hereinafter.
REFERENCES:
patent: 4890160 (1989-12-01), Thomas
patent: 4941044 (1990-07-01), Trew
patent: 4985767 (1991-01-01), Haghiri et al.
patent: 5008746 (1991-04-01), Bernard et al.
patent: 5068727 (1991-11-01), Haghiri et al.
patent: 5193004 (1993-03-01), Wang et al.
patent: 5253056 (1993-10-01), Puri et al.
patent: 5347308 (1994-09-01), Wai
patent: 5347312 (1994-09-01), Saunders et al.
patent: 5473383 (1995-12-01), Sezan et al.
patent: 6151363 (2000-11-01), Burl et al.
patent: 6289051 (2001-09-01), Bellers et al.
patent: 0630157 (1994-12-01), None
patent: WO9305616 (1993-03-01), None
Bellers Erwin B.
De Haan Gerard
Goodman Edward W.
Koninklijke Philips Electronics , N.V.
Lee Richard
LandOfFree
Motion estimation does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Motion estimation, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Motion estimation will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2826637