JPH0213078A - System and device for detecting moving vector - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

Detailed Description of the Invention (Industrial Application Field) The present invention relates to a method of transmitting a moving image signal by reducing the number of frames, such as a video conference or a video telephone, and particularly relates to a method for transmitting a moving image signal by reducing the number of frames, such as a video conference or a video telephone. The present invention relates to a method for detecting motion and performing motion compensation using this detection.

(Prior Art) There are many low-speed video communication services such as video conferences and video calls that transmit video signals by reducing the number of frames. Such video transmission systems use motion compensation coding and frame interpolation using motion vectors, so
It is important to detect high-granularity motion vectors from video signals.

In this case, since the frames of the moving image signal to be transmitted are thinned out, it is necessary to detect a motion vector not between adjacent frames as usual, but between frames separated by a plurality of frames.

For this reason, in the past, motion vectors were detected between adjacent frames in the original signal of the plane to be thinned out, and the motion vectors at the same spatial position in each frame were added together as shown in FIG. It was assumed that this is a motion vector between adjacent frames in an image signal with a reduced number of frames. In FIG. 3, F i-1+ F i +
Fill 1...F, +. -2゜F・・nsu・
1...F...10+ (i is any positive integer, n
is a positive integer representing the frame thinning rate) represents the frames of the original moving image signal, among which Fl and Fl. . Frames excluding Fl and F1+ are deleted. only that which is transmitted. This structure is repeated every n frames, and the number of frames of the moving image signal is reduced to 1/n.

(Problem to be Solved by the Invention) However, in the prior art, there are few problems when the motion vectors to be added are small, but when the motion vectors become large, motion vectors at the same spatial location in each frame are added together. Since the motion vectors that are added are not related to each other, it is difficult to accurately detect motion vectors between frames that are widely separated in time.

An object of the present invention is to detect a motion vector with high accuracy even in the above-mentioned cases.

(Means for Solving Problem J) In order to achieve this L1 goal, the present invention first detects the motion vector between adjacent frames from the source image signal of the surface whose number of frames is to be reduced, and Determine the position of one frame of the driving image signal shifted backward by the detector vector at that position from the corresponding position of the frame of the reduced moving image signal, and from this position, use the same method to determine the position of the driving image signal. Furthermore, determine the position one frame before,
This is repeated until the previous frame in the video signal with the reduced number of frames is reached, the motion vectors selected by this operation are added vectorially, and this is added to the video signal with the reduced number of frames. It is characterized in that it is a motion vector between frames.

(Principle of the Invention) The present invention will be explained in detail using FIG. In FIG. 1, F j−1+F i+F i+++..., Fin
,,-2*Fi+n-1+F++. ,F. . Ya, (i
is an arbitrary positive integer, and n is a positive integer representing the frame thinning rate) represents the frame of the driving image signal, among which F1 and F. . The frames except F, and F are deleted.
1+. shall be transmitted only. This structure is repeated every n frames, and the number of frames of the moving image signal is reduced to 1/n.

In the present invention, first, a motion vector is detected between adjacent frames F and Fk++ (k is any positive integer). In the present invention, there is no restriction on the method of detecting a motion vector between adjacent frames, and for example, the method described in "Image Motion Amount Detection Method" (Japanese Patent Application No. 13501/1982) can be used. Furthermore, there is no limit to the unit of motion vector detection; for example, a method using pixels as a unit, or a method of dividing a video frame into pixel blocks of a certain size and detecting a motion vector in units of pixel blocks, etc. may be used. be able to. For the sake of explanation hereinafter, it will be assumed that the motion vector is detected for each pixel block, but it is clear that other methods are not excluded.

In FIG. 1, the frame F1. The motion vector V between the upper pixel block B and the previous frame F in the video signal whose number of frames has been reduced is determined as follows.

Frame F1+ obtained before pixel block B. The frame F1+ is shifted by −v (i) from the center of the pixel block B, with the motion vector between the pixel block B and the pixel block B being V(1). -3
Let the pixel block located at the top position be B (1). At this time, -V (t) does not necessarily have to indicate the center of B (1).

Next, frame F1 obtained before pixel block B (1)
+. -2 is the motion vector (2), and the pixel block at the position on frame Fi++1-2 shifted by -V(2) from the center of pixel block B(1) is B
(2). Similar procedures are repeated until pixel block B (b) of the frame memory is determined.

Motion vector V(t) obtained by the above operation.

V (2) , ---, V (n-1) , V (n)
are added for each vertical and horizontal component, and this is added to frame F,
+. Let it be the motion vector V between the upper pixel block B and the previous frame F3 in the video signal whose number of frames has been reduced.

The relationship between the motion vectors added by the method described above is strengthened, and it is possible to accurately detect motion vectors between frames that are widely separated in time in a motion picture signal with a reduced number of frames.

In addition, the amount of processing required for motion vector detection does not increase, so
Easy to implement.

(Example) FIG. 2 shows an example according to the present invention. The driving image signal ip before the number of frames is reduced is first stored in the frame memory F
One image frame is stored in MI.

When the next image frame is input, this stored signal is transferred to the frame memory FM2 as a transfer image signal p. That is, the image signal of one frame is stored in F) 42 in terms of time.

Which image frame is to be transmitted is controlled by the input timing signal i, and based on this, the timing controller TC provides information L1 of the position where motion vector detection is to be performed to the motion vector detector MD. MD is FMI.

A motion vector dv at the position is detected using the image signal of FM2, and is provided to the timing determining circuit TT and vector adder VA.

From the TT head 1 and dv, the position 2 of the motion vector to be added to dv one frame before in the original signal is calculated and provided to the motion vector memory MMI. The MMI is a memory that stores motion vectors for one image frame, and contains the motion vector of one frame before in time.

MMI sets the motion vector pv at the relevant position to VA according to L2.
, VA adds dv and pv vectorially, and c
It is given to the motion vector memory MM2 as v. MM2 is t
cv is stored in the corresponding position according to l. in this way,
The latest motion vector information updated for each frame is stored in MM2, but once motion vector detection for one frame is completed, the contents of MM2 are transferred to MMI as motion vector t.
transferred as v.

The above processing is repeated for each image frame, and when the processing for each frame of the image signal whose number of frames has been reduced is completed, the contents of MM2 are outputted as the final motion vector 0L11 according to the readout signal rL.

Thereafter, the contents of MMI and MM2 are reset by the TC, and the same process is repeated thereafter.

(Effects of the Invention) As described above, the present invention calculates the motion vector between adjacent frames detected in the original moving image signal before the number of frames is reduced to the corresponding position of the frame of the moving image signal whose number of frames has been reduced. , by vectorially tracing in the reverse direction in time, adding the selected motion vectors vectorially, and using this as the motion vector between frames of the video signal whose number of frames has been reduced. , the granularity of motion vectors can be greatly increased.

Therefore, the present invention detects a highly turbulent motion vector from a moving image signal and greatly improves motion compensation efficiency, etc. in a method of transmitting a moving image signal by reducing the number of frames, such as in a video conference or a video telephone. It can be done and the effect is great.

[Brief explanation of the drawing]

FIG. 1 is a diagram explaining the present invention in detail, FIG. 2 is a diagram of an embodiment of the present invention, and FIG. 3 is a diagram explaining the principle of a conventional example. fiMl...Frame memory 1, FM2...Frame memory 2, TC...Timing controller, MD...Motion vector detector, TT...Timing determination circuit, MMI...Motion vector memory 1, VA...Vector adder, MM2...Moving vector memory 2゜

Claims (2)

[Claims] (1) In a method of transmitting a moving image signal by reducing the number of frames, the motion vector between each frame is detected for each position within the frame from the original moving image signal before the number of frames is reduced, and the number of frames is reduced. determining a position one frame before in the original moving image signal by shifting the detector vector of the position backwards from the position in the frame in the moving image signal obtained by moving the image signal; Determine the position of the previous frame in the original moving image signal, repeat the above operation until reaching the previous frame in the moving image signal with the reduced number of frames, and convert the motion vector detected by the above operation into a vector. , and the added motion vector is used as an inter-frame motion vector at the position within the frame in a moving image signal whose number of frames has been reduced. (2) A frame memory (1) that stores one image frame of the driving image signal before the number of frames is reduced, and a frame memory (1) that stores the next image frame when it is input. a frame memory (2) that receives image frames to be transferred; a timing controller that is controlled by an input timing signal that instructs which image frame is to be transmitted and outputs information on a position where motion vector detection is performed based on the input timing signal; Information on the position where motion vector detection is performed from the timing controller, and the frame memory 1 and the frame memory 2.
a motion vector detector that detects a motion vector at the position using an image signal of the motion vector; a timing determining circuit that calculates the position of the power motion vector; a motion vector memory 1 that receives the position of the motion vector to be added from the timing determining circuit and stores motion vectors for one image frame; a vector adder that receives motion vector information from one frame before and a motion vector at the position from the motion vector detector and adds them vectorially; and a position where motion vector detection is performed from the output of the vector adder and the timing controller. 1. A motion vector detection device comprising a motion vector memory 2 that receives information on the motion vector and stores the latest motion vector information updated for each frame.