JP3001753B2 - Motion vector detection method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for detecting a motion vector in a high-efficiency coding method for a moving image.

[0002]

2. Description of the Related Art As a high-efficiency coding method for a moving image, a predictive coding method for estimating a motion vector between frames or between fields for each block and performing motion compensation has been used. In particular, in inter-frame / inter-field adaptive prediction in which a motion vector between frame signals and a motion vector between field signals are adaptively switched and used in order to perform efficient code compression, both inter-frame and inter-field prediction are performed. Needs to detect the motion vector of

Conventionally, when detecting a motion vector between a field and a frame for each block, a motion vector is individually obtained.

[0004]

In the conventional method, when a motion vector between frames is detected, a search process is performed in the same search range as that for detecting a motion vector between fields. Was needed.

SUMMARY OF THE INVENTION An object of the present invention is to solve the above-mentioned problems and simplify the process of searching for a motion vector between frames by using a motion vector between fields as an initial value in detecting a motion vector between frames. It is in.

[0006]

According to the present invention, the inter-field motion vector is determined, and then the inter-field motion vector is set as an initial value, and a search is performed in a small area of an arbitrary shape centered on the inter-field motion vector. Is obtained.

In addition, a motion vector between frames is obtained by performing a search in a rectangular area of m × n pixels centered on the inter-field motion vector as an initial value.

FIG. 1 is a diagram for explaining the principle of the present invention.
First, an inter-field motion vector is detected. As shown in FIG. 1A, the inter-field motion vector includes (1) the first field 101 of the current frame 100 and the reference frame 2
00 first field 201, (2) current frame 100
, The first field 101 and the second field 202 of the reference frame 200, (3) the second field 102 of the current frame 100 and the first field 20 of the reference frame 200
1, (4) Four types of the second field 102 of the current frame 100 and the second field 202 of the reference frame 200 are detected.

The initial value when detecting the inter-frame motion vector is
Average value, median value,
Using a representative value or the like, an inter-frame motion vector is detected using a small area of an arbitrary shape centered on the inter-field motion vector or an m × n rectangular area as a search range. That is, as shown in FIG. 1B, when the inter-frame motion vector is detected by the matching error calculation from the block 103 of the current frame 100 and the corresponding block 203 of the reference frame 200, the inter-field motion vector is centered. Then, the block 203 is moved by a predetermined range 300 to obtain a position where the matching error is minimized, and an inter-frame motion vector is detected.

[0010]

[Function] The inter-field motion vector and inter-frame motion vector in an interlaced image have a very high correlation, and the inter-frame motion vector search can obtain most of the motion only by searching a small area around the inter-field motion vector. it can. Since the computational complexity of the motion vector search is greatly affected by the search range, the search range at the time of detecting the inter-frame motion vector is narrowed by using the inter-field motion vector obtained earlier as the initial value. The amount of calculation at the time of detecting a motion vector can be reduced.

[0011]

FIG. 2 shows an example of a motion vector detecting circuit embodying the present invention. In FIG. 2, 1 is an interlace signal of the current frame, 2 is a field memory, 3 is a first field signal of the current frame, 4 is a field memory, 5 is an interlace signal of a reference frame, 6 is a field memory, and 7 is a reference. The first field signal 8 of the frame represents a field memory.

9 is an address signal for reading the second field of the current frame, 10 is a second field signal of the current frame, 11 is an address signal for reading the first field of the current frame, 12 is a first field signal of the current frame, and 13 Is an address signal for reading the second field of the reference frame, 14 is a second field signal of the reference frame, 15 is an address signal for reading the first field of the reference frame, 16 is a first field signal of the reference frame,
17 is a small area divider for the second field of the current frame, 1
8 is a small area divider for the first field of the current frame, 19
Represents a small area divider of the second field of the reference frame, and 20 represents a small area divider of the first field of the reference frame.

Reference numeral 21 denotes a current frame second field block signal, 22 denotes a current frame first field block signal, 23 denotes a reference frame second field block signal, 24 denotes a reference frame first field block signal, and 25 denotes a current frame. A frame signal generator, 26 is a frame signal generator for a reference frame, 27 is a current frame block signal, 28 is a reference frame block signal, 2
9 is a matching error calculation unit, 30 is a difference value, 31 is a comparator, 32 is a memory for storing a block difference minimum value, 33
Is the address data, 34 is the address memory, 35 is the address data of the inter-field motion vector, 36 is the arithmetic unit,
Reference numeral 37 denotes address data which is a reference value for detecting an inter-frame motion vector.

The interlace signal 1 of the current frame is input, and the first field and the second field are stored in the field memory 4 and the field memory 2, respectively. Similarly, the first and second fields of the interlaced signal 5 of the reference frame are also stored in the field memories 8 and 6, respectively.

If the first field of the current frame is (x,
The value of the address signal 11 consisting of the two coordinates y) is read out from the field memory 4 as an offset value, and is input to the small area divider 18. The small area divider 18 divides the first field of the current frame into blocks and outputs a current frame first field block signal 22. Similarly, the first field of the reference frame is the small area divider 2
It is divided by 0 and output as the first field block signal 24 of the reference frame. The block signal 22 and the block signal 24 are input to the matching error calculator 29.

The matching error calculator 29 outputs a difference value 30 between the input block signals 22 and 24 to a comparator 31. The comparator 31 has a minimum value memory 32,
If the input difference value 30 is smaller than the value stored in the memory 32, the value of the memory 32 is updated, and the two coordinates (x,
The address 33 consisting of y) is recorded in the address memory 34. After the above operation is completed within the motion vector search range, the motion vector between the current frame first field and the reference frame first field is stored in the address memory 34.

Hereinafter, similarly, the output 22 of the small area divider 18
The motion vector between the first field of the current frame and the second field of the reference frame is obtained by the output 21 of the small area divider 17 and the output 23 of the small area
0, the motion vector between the second field of the current frame and the first field of the reference frame is determined by the output 21 of the small area divider 17 and the output 23 of the small area divider 19. All motion vectors between the second fields are stored in the address memory 34.

Next, the first field block signal 22 and the second field block signal 21 of the current frame
Is input to the frame signal generator 25. In the frame signal generator 25, two field block signals 2
1 and 22 are combined and the current frame block signal 27 is output. Similarly, a reference frame block signal 28 is created from the field block signals 23 and 24 of the reference frame.
The offset values read from the field memories 6 and 8 of the reference frame need to be shifted by the reference value calculated by the inter-field motion vector. For this reason, for the motion vector used as the reference value for detecting the inter-frame motion vector among the four detected inter-field motion vectors, a predetermined operation such as an average value, a median value, a representative value, etc. Done at
Input to the small area dividers 19 and 20.

The address signals 13 and 15 from the small area dividers 19 and 20 correspond to (x, y) output values 3
7 is added. The first and second fields of the reference frame read by the address signals 13 and 15 are:
The signal is divided into blocks by the small area dividers 19 and 20, and is output as a reference frame block signal 28 by the frame signal generator 26. As in the case of detecting the inter-field motion vector, the inter-block difference value 30 is calculated by the matching error calculation unit 29, and the address 33 having the minimum value is detected by the comparator 31. Here, the small area divider 1
The blocks divided by 9 and 20 are moved in the range of a small area of an arbitrary shape centering on the pixel position obtained by the motion vector between fields, and an address having the minimum value is searched in that range. If this search range is a rectangular area of m × n (m and n are predetermined natural numbers), the small area dividers 19 and 2
Address generation control for searching by 0 is simplified.
For example, a pixel position obtained by a motion vector between fields, that is, an (x, y) output value 37 from the arithmetic unit 36
± m / 2 pixels in the row direction and ± n /
The reading of blocks for calculating a matching error is repeated by using two lines as a search range, and a motion vector between frames is detected. Finally, the inter-frame motion vector is stored in the address memory 34.

The example shown in FIG. 2 is an example in which the full search block matching method is used to search for inter-field motion vectors. However, the inter-field motion vector search can be performed by any method such as a hierarchical motion vector search method or a tree search. Can be used.

[0021]

As described above, according to the present invention,
By using the inter-field motion vector as the initial value, the search range at the time of detecting the inter-frame motion vector is narrowed, so that the amount of calculation required to detect the inter-frame motion vector can be reduced.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating the principle of the present invention.

FIG. 2 is a diagram illustrating an example of a motion vector detection circuit embodying the present invention.

[Explanation of symbols]

 101 First field of current frame 102 Second field of current frame 103 Block 201 First field of reference frame 202 Second field of reference frame 203 Block 300 Range

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-6-181568 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) H04N ^7/ 24-7/68

Claims (2)

(57) [Claims] [Claim 1] among the motion compensation method in an image coding method for compressing the code amount of interlaced image, the continuous
In a motion vector detection method in inter-frame / inter-field adaptive prediction in which a motion vector between frame signals obtained by merging first and second field signals and a motion vector between field signals are adaptively switched and used, some frame signals are detected. When calculating the inter-field motion vector and inter-frame motion vector for each block , the first frame of the current frame is used.
Field and first field of reference frame, current frame
The first field of the reference frame and the second field of the reference frame.
The second field of the current frame and the first field of the reference frame
Field and the second and reference fields of the current frame.
Four types of movement between the fields of the second field of the frame
Mean, median, or
A motion vector comprising detecting a motion vector between frames using a representative value as a search range with at least an area smaller than the entire area of the frame of an arbitrary shape centered on a pixel position obtained by the inter-field motion vector. Detection method. Wherein among the motion compensation method in an image coding method for compressing the code amount of interlace image, the continuous
In a method for detecting inter-frame / inter-field adaptive prediction using adaptively switching between a motion vector between frame signals obtained by merging the first and second field signals and a motion vector between field signals, some frame signals are used. When calculating the inter-field motion vector and inter-frame motion vector for each block , the first frame of the current frame is used.
Field and first field of reference frame, current frame
The first field of the reference frame and the second field of the reference frame.
The second field of the current frame and the first field of the reference frame
Field and the second and reference fields of the current frame.
Four types of movement between the fields of the second field of the frame
Mean, median, or
Using a representative value, a rectangular area of m × n (m, n is an integer of 2 or more and smaller than the number of horizontal and vertical pixels of a frame) pixels centered on a pixel position obtained by a motion vector between the fields is searched. A motion vector detection method, comprising detecting a motion vector between frames as a range.