Television – Image signal processing circuitry specific to television – Motion vector generation
Patent
1996-11-08
1999-09-14
Lee, Richard
Television
Image signal processing circuitry specific to television
Motion vector generation
348420, H04N 732
Patent
active
059530790
DESCRIPTION:
BRIEF SUMMARY
FIELD OF THE INVENTION
This invention relates to video image motion measurement and in particular to such motion measurement when there is a camera zoom, rotation, or some other non-linear translation between adjacent fields of the video image.
BACKGROUND OF THE INVENTION
In our United Kingdom Patent No. GB-B-2188510 a method for TV picture motion measurement is described in which a plurality of motion vectors are derived for each of a plurality of measurement blocks of a picture and from these a list of motion vectors is derived which are applicable over the whole area of each block. The process is then repeated over the whole area of the picture. These vectors are derived using a phase correlation technique. One of these is then assigned to each elementary area of each block of the image. Other methods are also possible for generating such a list of motion vectors, for example the block matching technique described in European Patent Application No. EP-A-0395264. Each elementary area of a region of the picture to which a vector is assigned may be as small as a picture element (pixel) or it may comprise a plurality of picture elements for a block of the picture. The motion vectors thus derived may be used, for example, to generate output fields at instants in time intermediate of the time of two input fields. Such intermediate fields are required when producing slow motion effects, when transferring video images to or from film, or in standards conversion.
The methods used for deriving the motion vectors have been found to be highly satisfactory for stationary camera positions but have led to problems when a zoom of the camera is involved or a rotation of the camera about its viewing access. These problems have been particularly serious when such a zoom or rotation has been combined with a camera pan and the methods for deriving the motion vectors have in some cases broken down completely in these situations.
SUMMARY OF THE INVENTION
Preferred embodiments of the present invention seek to provide a machine method for extracting a motion component due to camera zoom or rotation from the gross movements in a scene. The component for each picture region will be proportional to the distance of that region from a particular point in the image. This is achieved by estimating a constant representing a non-linear transformation, from sets of trial vectors and then adding a vector defining the non-linear transformation to the trial vector list for each block.
The invention is defined in the appended claims to which reference should now be made.
BRIEF DESCRIPTION OF THE DRAWINGS
Preferred embodiments of the invention will now be described in detail by way of example with reference to the accompanying drawings in which:
FIG. 1 shows schematically the type of motion vector derived by a conventional TV picture motion measurement system;
FIG. 2 shows schematically the effect of a camera pan;
FIG. 3 shows the effect of a camera zoom on the top row of picture regions shown in FIG. 2;
FIG. 4 and 5 show zoom and pan surfaces derived using an embodiment of the invention;
FIG. 6 shows a block diagram of an apparatus according to a preferred embodiment of the present invention.
FIG. 7 is a schematic illustration of a process for extracting a motion component due to camera zoom in accord with a preferred embodiment of the present invention; and
FIG. 8 is a schematic illustration of a simple process for extracting a motion component due to camera zoom in accord with a preferred embodiment of the present invention.
DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS
In FIG. 1 two adjacent frames of a television signal, F0 and F1, are shown schematically. These are separated by a time T. An object 2 is shown in both F0 and F1 and this object moves from a first position in F0 to a second position in F1. A TV picture motion measurement system of the type described in GB2188510 is used to analyse differences between F0 and F1 and this produces a vector V which describes the motion of the object 2 between the adjacent frames or
REFERENCES:
patent: 4677476 (1987-06-01), Kondo
patent: 5008746 (1991-04-01), Bernard et al.
patent: 5083860 (1992-01-01), Miyatake et al.
patent: 5247355 (1993-09-01), Frederiksen
patent: 5259040 (1993-11-01), Hanna
patent: 5267034 (1993-11-01), Miyatake et al.
patent: 5357287 (1994-10-01), Koo et al.
patent: 5371539 (1994-12-01), Okino et al.
patent: 5376971 (1994-12-01), Kadono et al.
M. Hoetter, "Differential Estimation of the Global Motion Parameters Zoom and Pan", Signal Processing, vol. 16, No. 3, Mar. 1989, pp. 249-265.
Burl Michael
Thomson Roderick
British Broadcasting Corporation
Lee Richard
Neuner George W.
LandOfFree
Machine method for compensating for non-linear picture transform does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Machine method for compensating for non-linear picture transform, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Machine method for compensating for non-linear picture transform will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-1514987