JP2898787B2 - Motion vector detection device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image motion vector detecting apparatus using a composite gradient method.

[0002]

2. Description of the Related Art In a conventional apparatus for detecting a motion vector of an image using a composite gradient method, a motion vector can be generally described as follows.

[0003]

(Equation 1) That, Delta] t _i, [Delta] x _i, each field or inter-frame difference signal [Delta] y _i, horizontal gradient, the detected vector v and vertical gradient (v _x, v _y) is

[0004]

(Equation 2) Becomes Here, i is a non-zero positive integer.

Here, the compound of the compound gradient method means that when a screen is divided into several small blocks and a motion vector of an image is obtained for each block, a plurality of pixels i belonging to the small block are considered. It is named in the sense that the motion vector of the small block is obtained from the field or inter-frame difference signal by using the calculation method of Expression (1) or Expression (2), respectively. Also, | Δx of the denominator of equation (2)
_i | and | [Delta] y _i | is, x and y directions, respectively, that is to say the absolute value of the slope signal determined by the difference in the horizontal scanning direction and the vertical direction adjacent-pixel image signals of the screen.

[0006]

As is clear from the equations (1) and (2) described in the section of the prior art, in the conventional compound gradient method, in a region where the gradient signal is small, the denominator of the arithmetic expression, that is, the gradient is used. The sum of absolute values of the signal becomes a value close to zero. Therefore, the value of the detection vector shows a very large value as compared with the true vector value, and a large error occurs. Accordingly, an object of the present invention is to provide an image motion vector detecting device which eliminates the above-mentioned disadvantages.

[0007]

In order to achieve the above object, a motion vector detecting apparatus according to the present invention divides a screen into a plurality of small blocks, and for each block, moves a motion vector of a representative image horizontally and vertically. A motion vector detection device obtained by a composite gradient method as a set of components, and
A motion vector estimating device that obtains a horizontal or vertical component of the motion vector by the composite gradient method adds a field or inter-frame difference value Δt _i obtained for each of a plurality of positive integer i pixels constituting the small block to the motion vector detecting device. A multiplier multiplied by a horizontal gradient code sign (Δx _i ) or a vertical gradient code sign (Δy _i ) considering the sign of the spatial gradient corresponding to each position of the difference value Δt _i , and i obtained by the multiplier A first integrator for integrating the multiplied values, i.
Absolute value of the horizontal gradient corresponding to each of the plurality of pixels | Δ
a second integrator that integrates x _i | or the absolute value of the vertical gradient | Δy _i | and a division that divides the integrated value obtained by the first integrator by the integrated value obtained by the second integrator A motion vector detection device comprising a divider and a horizontal or vertical component of a motion vector obtained from the divider as a horizontal or vertical component, thereby obtaining a motion vector of the small block. Horizontal gradient Δx corresponding to
_i or the vertical gradient Δy _i to the dither signal S ₁ or S ₂
And the absolute value | Δx _i + S ₁ | or | Δy
_i + S ₂ | is taken and supplied as an input signal to the second integrator.

[0008]

According to the motion vector detecting apparatus of the present invention using the composite gradient method, the divisor Σ | Δx in the equations (1) and (2) is used.
_{Even if i} | and Σ | Δy _i | become small and approach zero indefinitely, since the small but finite dither signals S ₁ and S ₂ are added to Δx _i and Δy _i , Equation (1) ) And the divisor of equation (2) are finite, so that no large error occurs in the motion vector.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in detail below with reference to the accompanying drawings. FIG. 2 is a block diagram for explaining the conventional compound gradient method. Signs Δt _i , Δx _i , Δy _i
Is the difference between the fields of the pixel of interest, the horizontal gradient, and the vertical gradient, the detection vector v (v _x , v _y ) is expressed by the following equation as described in the section of the prior art.

[0010]

(Equation 3) On the other hand, FIG. 1 shows a block diagram of a motion vector detecting device using the composite gradient method according to the present invention. As shown in FIG. 1, dither signals S ₁ and S ₂ are added to the gradient signal.
The detection vector v (v _x , v _y ) can be expressed as follows.

[0011]

(Equation 4) As shown in FIG. 1, since the dither signals S ₁ and S ₂ are AC signals having an average value of zero, the effect of the addition of the dither signal on the output of the integrator 6 is large in a region where the gradient signals Δx _i and Δy _i are large. Although zero comb is a value very close to zero, as the gradient signal becomes smaller, the effect of adding the dither signals S ₁ and S _{2 on} the output of the integrator 6 becomes relatively larger.

That is, the absolute values of v _x and v _y in Equations (3) and (4) are not affected by dither addition in a region where the gradient is large, and are expressed by Equation (2) in a region where the gradient is small. By increasing the denominator, a value smaller than Expression (2) is indicated. The method of adding a dither signal to this gradient signal is more effective in detecting a motion vector by a gradient method using an initial deviation vector (see FIG. 3). It is assumed that a value close to the true motion vector is selected as the initial deviation vector. In the method in which dither addition is not performed on the gradient signal, the absolute values of the displacement vectors v _x and v _y obtained from the equation (2) become very large in a region where the gradient is small, and the detection vector becomes significantly wrong. Become. On the contrary,
In the method in which the dither signal is added to the gradient signal, even in a region where the gradient is small, v _x and v _y are close to zero, so that the detection vector has a value close to the initial deviation vector, and does not cause a large error.

[0013]

According to the conventional composite gradient method, the detected vector value shows an extremely large value as compared with the true vector value in an area where the gradient signal is small, and becomes an error vector. In the present invention, the addition of the dither signal to the gradient signal does not cause a large error in the detection vector even in the region where the gradient signal is small, and has no effect in the region where the gradient signal is large. When the present invention is applied to the conventional composite gradient method using the initial deviation vector, the deviation vector in a region where the gradient signal is small due to the influence of the dither signal is close to zero, and the detection vector is the initial deviation vector. , And if the initial displacement vector is correct, a correct value is calculated for the motion detection vector.

[Brief description of the drawings]

FIG. 1 shows a block diagram of a motion vector detection device according to the present invention.

FIG. 2 shows a block diagram of a conventional motion vector detection device.

FIG. 3 is a block diagram in which the present invention is applied to a conventionally used gradient method using an initial deviation.

(72) Inventor Shoichi Suzuki 1-10-11 Kinuta, Setagaya-ku, Tokyo Japan Broadcasting Corporation Research Institute (72) Inventor Gen Sonehara 1-110-11 Kinuta, Setagaya-ku, Tokyo Japan Broadcasting Japan Broadcasting Corporation Research Institute (72) Inventor Junji Kumada 1-10-11 Kinuta, Setagaya-ku, Tokyo Japan Broadcasting Corporation Research Institute (58) Field surveyed (Int.Cl. ⁶ , DB name) H04N 7/24 -7/68 G06T 7/20

Claims (1)

(57) [Claims] 1. A motion vector detecting device which divides a screen into a plurality of small blocks and obtains a motion vector of an image representing each block as a set of horizontal and vertical components by a composite gradient method, and A motion vector estimating device that obtains a horizontal or vertical component of the motion vector by the composite gradient method adds a field or inter-frame difference value Δt _i obtained for each of a plurality of positive integer i pixels constituting the small block to the motion vector detecting device. Horizontal gradient code sign (Δx _i ) or vertical gradient code sign (Δy _i ) considering the sign of the spatial gradient corresponding to each position of the difference value Δt _i
, A first integrator for integrating the i multiplied values obtained by the multiplier, and an absolute value | Δx _i | or a vertical gradient of a horizontal gradient corresponding to each of the i pixels. A second integrator that integrates the absolute value | Δy _i | of the first and second integrators, and a divider that divides the integrated value obtained by the first integrator by the integrated value obtained by the second integrator. In a motion vector detecting apparatus which obtains a motion vector of the small block from the divided value obtained from the divider as a horizontal or vertical component of the motion vector, a horizontal gradient Δx corresponding to each of the plurality of pixels is provided. _i or the vertical gradient Δy _i plus the dither signal S ₁ or S ₂ , and then the absolute value | Δx _i + S ₁ | or | Δy _i + S ₂
| Is supplied as an input signal to a second integrator.