JP3137084B2 - Motion detection apparatus and 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 and a device for detecting a motion amount, and more particularly to a motion amount detecting device and a method for detecting a motion amount from two images composed of digital image signals.

[0002]

2. Description of the Related Art With the use of moving images and the progress of signal processing relating to moving images, the need for a technique for detecting the direction and magnitude of the motion of each object in an image, that is, a motion vector, has been increasing. For example, motion-compensated image coding for high-efficiency moving image compression, moving object detection processing in a traffic monitoring system or a visual sensor of an autonomous vehicle, speed detection processing, and the like.

Also, in an image display apparatus based on a subfield system such as a plasma display panel, in order to reduce moving image noise peculiar to moving images, a motion of an image is detected and a signal processing for noise suppression using the detected motion is performed. Is effective.

A series of images taken in close proximity in time,
Alternatively, as a method of determining a motion vector from two images, a gradient method of determining a motion vector from a spatial gradient of luminance and a temporal change of luminance, and the other image while moving each block of a blocked image Or a correlation method that determines the motion vector from the amount of movement that maximizes it, or calculates the difference between images while moving the image and determines the motion vector from the amount of movement that minimizes the difference A pattern matching method and the like are known. Also,
When calculating the correlation between images in the correlation method, a technique for performing processing at high speed using fast Fourier transform (FFT) is also well known.

Further, a phase correlation method for determining a motion vector from a peak value of a phase correlation function using a fast Fourier transform is also known.

[0006]

However, the correlation method,
The phase correlation method has a problem that a motion vector cannot be correctly detected when a change in a luminance signal is unclear. In order to solve this problem, for example, Japanese Patent Application Laid-Open
Japanese Patent No. 6989 discloses that an edge portion in an input image is detected by an edge detecting means, an image is sharpened only for the detected edge region, and then a motion vector is detected using a phase correlation method. There has been proposed a motion amount detection device that performs the operation.

As described above, the correlation method or the phase correlation method is widely used because the processing load is relatively light, but the spatial change of the luminance signal is smooth and unclear as described above. In this case, there is a problem that a motion vector cannot be correctly detected. This will be described in detail below.

[0008] For example, consider a case in which the luminance changes spatially gently and monotonously as shown in FIG. Originally, the image signal should be represented on a two-dimensional coordinate axis, but for simplicity of description, FIG. 5 shows a signal in a one-dimensional case.

In the correlation method, signal overlap is calculated while moving the signal block f in order to detect the position of the signal block f in the range indicated by the arrow in FIG. As a result, a correlation function as shown in FIG. 6 is obtained. However, the correlation function does not reach the maximum at the correct position (in this case, the position 64), and the correlation increases as the luminance increases. You can see that it goes.

[0010] Therefore, in such a case, the correlation method cannot determine a correct motion amount.

As another problem, the correlation method has a problem that when there is a region having a remarkably high luminance in an image due to reflected light or the like, a motion vector cannot be correctly determined due to the influence of the region.

For example, let us consider a case where an extremely high luminance region is isolated in an image as shown in FIG. When the correlation function is calculated based on the correlation method, FIG.
Is obtained, and the correlation function does not reach the maximum at the correct position (again, the position 64) due to the influence of the high luminance area. Therefore, also in this case, it can be seen that the motion amount cannot be determined correctly.

The above-mentioned Japanese Patent Application Laid-Open No. 7-66989 proposes a method for detecting a motion vector even for an image in which the change in luminance signal is unclear. It is required that a rapidly changing edge portion exists in an image block whose motion vector is to be determined. For this reason, as shown in FIG. 5, a motion vector cannot be correctly determined in a region where the luminance changes gradually.

In order to analyze the motion of an image or to suppress the noise of a moving image in a subfield type image display device, it is important to detect a motion vector even in a region where the luminance changes gradually.

Therefore, the present invention has been made in view of the above-mentioned problems, and has as its object to solve the problem even in a region where the luminance is gradually changing or when there is a region where the luminance is extremely high in the image. Another object of the present invention is to provide a motion detection method and apparatus capable of detecting a motion vector with high accuracy.

[0016]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention provides a luminance range setting means for setting a plurality of luminance ranges, and only a pixel having a luminance within the luminance range according to each of the luminance ranges. An image decomposing unit that extracts from the input image and decomposes the input image into a plurality of partial images, a correlation function calculating unit that calculates a correlation function for each of the decomposed partial images, and a correlation function for each of the partial images. Motion vector determining means for determining a motion vector in total.

According to the present invention, there is provided a brightness range setting means for setting a plurality of brightness ranges, and extracting only pixels having brightness within the brightness range from an input image in accordance with each of the brightness ranges, and dividing the input image into a plurality of portions. Image decomposing means for decomposing into images, phase correlation function calculating means for calculating a phase correlation function for each of the decomposed partial images, and a motion vector for determining a motion vector by integrating the phase correlation functions for each of the partial images Determining means.

According to the present invention, there is provided a brightness range setting means for setting a plurality of brightness ranges, and extracting only pixels having brightness within the brightness range from an input image in accordance with each of the brightness ranges, and dividing the input image into a plurality of portions. Image decomposing means for decomposing into images, motion vector calculating means for calculating a motion vector for each of the decomposed partial images, and determining the final motion vector by integrating the motion vectors calculated for each of the partial images Motion vector determining means.

In the present invention, the luminance range setting means may adaptively variably set a luminance range according to the input image.

[0020]

Embodiments of the present invention will be described. The present invention, in one of its preferred embodiments,
Referring to FIG. 1, in an image decomposing unit (2) that receives two blocked input images as inputs, a plurality of luminance ranges set by the luminance range setting unit (1) have luminance within each luminance range. Pixels are respectively extracted from the input image, and a partial image having a luminance of zero is generated for pixels having other luminances, and each of the two input images is decomposed into a partial image. A plurality of luminance ranges (for example, when the luminance takes a value of 0 to 255, [0, 1
5], [16, 31], [240, 255], etc.), the image decomposing means (2).
Decomposes two input images into a plurality of partial images.

Next, in the correlation function calculating means (3), for each of the partial images corresponding to the same luminance range, the two divided input images decomposed by the image decomposing means (2) are correlated. The motion vector deciding means (4) calculates the sum of the correlation functions for each of the partial images and adds up the correlation functions for all of the partial images per input image (for the number of luminance ranges) to determine the pixel position at which the maximum value is obtained To determine a final motion vector.

According to the second embodiment of the present invention, referring to FIG. 2, the image dividing means (2) has a luminance within each luminance range for a plurality of luminance ranges with respect to two input images. By extracting pixels from the input image, the two input images are respectively decomposed into partial images corresponding to the number of the luminance ranges, and a phase correlation function calculating means (5)
Then, a phase correlation function for each of the divided partial images corresponding to the same luminance range is calculated between the two input images, and the motion vector determining means (4) calculates a phase correlation function for each of the partial images. The final motion vector is determined by detecting the pixel position at which the sum of the correlation functions added for all the partial images per input image has the maximum value.

Further, in the third embodiment, referring to FIG. 3, in the image dividing means (2), the two input images are divided into a plurality of luminance ranges according to the respective luminance ranges. Pixels having a luminance within the luminance range are respectively extracted from the input image, and the two input images are decomposed into partial images corresponding to the number of the luminance ranges, respectively. A motion vector is calculated for each partial image, and the motion vector determining means (4A)
The final motion vector may be determined by integrating the motion vectors calculated for each of the partial images. The motion vector calculation means (6) obtains a correlation function for each of the decomposed partial images corresponding to the same luminance range between the two input images, and detects a position where the correlation function takes a maximum value. Thus, a motion vector is calculated for each partial image, and the motion vector determining means (4A) determines the final motion vector by averaging the motion vectors for each partial image.

Further, as a fourth embodiment of the present invention, referring to FIG. 4, the luminance range may be adaptively variably set from an input image.

As described above, in the embodiment of the present invention, by calculating the correlation function for each of the images decomposed for each luminance range, the correlation value is calculated between the regions having greatly different luminance values. Correlation between regions that should not correspond does not disrupt the correlation function.

As a result, even in the case where there is a region having a remarkably high luminance in the input image, the motion vector can be correctly determined without being affected by the region.

For the same reason, a motion vector can be correctly determined even in a region where luminance changes gradually.

[0028]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an embodiment of the present invention;

[Embodiment 1] FIG. 1 is a diagram for explaining the configuration and processing of a first embodiment of the present invention. In the first embodiment of the present invention, a process of calculating a motion vector for each divided block from two digital images photographed close in time is performed. Of the two input digital images, the image f is assumed to be divided into blocks of a predetermined size, for example, a size of 16 × 16 pixels, and the image thus pre-blocked is input. And In the present embodiment, a motion vector is calculated by detecting, from another image g, a region where the same image as that of the block is located.

Assuming that the search range is, for example, 128.times.128 pixels centered on the position of this block, the image g also has a block of 128.times.1
It is assumed that the image is cut into 28 pixels and input.

The sizes of the two types of image blocks are set to different sizes here, but they may be the same. The luminance range of the input image is 0 to 255
And

First, the brightness range setting means 1 sets a brightness range to be used in the subsequent processing. Here, a fixed luminance range set in advance, for example, [0, 15],
[16, 31], [32, 47],. , [240,2
55] is stored for each of the 16 luminance ranges divided for each of 16 gradations.

The image decomposing means 2 decomposes the input images f and g into 16 partial images according to these 16 luminance ranges.

Here, the image f is represented by f (x, y), and the image g is represented by g
When expressed by (x, y), for example, the first luminance range, [0,
15] and partial images f1 (x, y) and g1
(X, y) is defined by the following equations (1) and (2).

[0035]

(Equation 1)

(Equation 2)

Partial image fj corresponding to another luminance range
(X, y), gj (x, y) (where j = 2 to 16)
Is generated by a similar rule. That is, for the given luminance range, if the luminance of the input image is within the luminance range, the value is given to each pixel, otherwise, the value zero is given to each pixel.

In this way, the image decomposing means 1
The input images f (x, y) and g (x, y) are respectively

[0038]

(Equation 3)

Then, the image is decomposed into a plurality of partial images, in this embodiment, 16 partial images.

Next, the correlation function calculating means 3 calculates a correlation function cj of the following equation (4) for each of the partial images thus decomposed.
Calculate (x, y).

[0041]

(Equation 4)

The calculation of this correlation function (convolution operation) is actually performed as follows using a fast Fourier transform in order to speed up the operation.

First, an image fj (size 16 × 16) and
In order to make the size of the image gj (size 128 × 128) uniform, the image fj is arranged at the center of the block of size 128 × 128 pixels, and the periphery is filled with zero.

The image thus obtained is represented by f'j (x, y)
(Size 128 × 128 pixels).

Next, the Fourier transforms Fj, G of f'j, gj
j is calculated using the fast Fourier transform.

Next, the product of the complex conjugate Fj * of Fj and Gj is calculated.

Finally, the correlation function cj (x, y) is obtained by calculating the inverse Fourier transform of the product, Fj * xGj, also using the fast Fourier transform (IFFT).

In order to avoid the effect of discontinuity at the boundary of the image gj (Gibbs phenomenon, etc.), when calculating the Fourier transform, a window function such as a 3 sigma-Gaussian window is multiplied, and the area around the boundary is multiplied. A method of calculating the Fourier transform after attenuating the value may be used.

The correlation function cj (x, y) thus obtained is
From the position at which it takes the maximum value (xjmax, yjma
x), a motion vector vj = (xjmax, yjma) obtained from the partial images fj, gj
x) is calculated, but in the present embodiment, this processing is not performed, and the correlation function cj itself is sent to the motion vector determination means 4.

Finally, the motion vector determining means 4 calculates the correlation function cj (j = 1-16) calculated for each partial image.
To determine a motion vector.

In this embodiment, the motion vector determining means 4
Calculates the sum C (x, y) by adding each correlation function.

[0052]

(Equation 5)

Next, the position (xmax, ymax) at which the sum C (x, y) takes the maximum value is detected.

Then, this vector is defined as a motion vector v =
(Xmax, ymax) is determined and output.

Here, the principle of motion vector detection by the correlation method will be considered. In the correlation method, the overlapping degree is calculated while shifting the image f to detect the position where the blocked image f has moved in the image g, and the best overlapping position is detected. That is, the correlation function C of the following equation (6)
0 (x, y) is calculated, and the position where the correlation function takes the maximum value is detected.

[0056]

(Equation 6)

Incidentally, the correlation function C0 (x, y)
Can be expressed as in the following equation (7) using the partial images fj and gj described above.

[0058]

(Equation 7)

In equation (7), the first term, C (x, y)
Is the sum of the correlation functions between the partial images calculated to determine the motion vector in the present embodiment. Second
The term represents the sum of mutual correlation functions between partial images belonging to different luminance ranges.

Incidentally, the partial image f belonging to the luminance range j
j is a partial image gj which also normally belongs to the same luminance range j
Therefore, the term that is originally effective for detecting the motion vector should be the first term C (x, y) in the above equation (7). And the second term is the first
The term blurs and acts as a hindrance for motion vector detection.

In this embodiment, the harmful second term is removed by decomposing the input image into partial images and calculating the correlation, and only the first term effective for detecting the motion vector is calculated. .

Then, as a result of detecting the motion vector using the first term of the above equation (7), more accurate motion vector detection becomes possible, and even in a region where the luminance is difficult to change, which was difficult in the past. In addition, even when an image has a region with extremely high luminance, a motion vector can be accurately detected.

FIGS. 5 to 10 show the results of comparison between this embodiment and a conventional simple correlation method as a comparative example. Although an image signal is originally represented by two-dimensional spatial coordinates, a comparison is made here for a one-dimensional signal in order to simplify the description. The two-dimensional case does not change the essential result.

FIG. 5 shows a signal block g in which the position of a signal block f (range indicated by an arrow) is searched. Since the center position of the signal block f is 64 in the signal block g, 64 is the answer to be obtained in this case.

FIG. 6 shows a block f obtained by a conventional correlation method.
And g are obtained while shifting the position of the block f. The horizontal axis represents the center position of the block f. It can be seen that the result does not reach the maximum at the position 64, and the correlation increases as the luminance shifts to the higher luminance.

That is, as shown in FIG. 5, in a region where the luminance changes gently and monotonously, the motion vector cannot be accurately calculated by the conventional correlation method.

FIG. 7 shows the result of the processing according to this embodiment. FIG. 7 shows a sum C of correlation functions between partial images belonging to the same luminance region. Referring to FIG. 7, it can be seen that the maximum value is obtained at the correct position 64.

As described above, according to the present embodiment, it can be seen that a motion vector can be correctly detected even in a region where the luminance changes gently and monotonously, which is difficult to detect by the conventional method.

8 to 10 show other examples. FIG.
Is a signal block g in which the position of the signal block f (the range indicated by the arrow) is searched. Again, the center position of the signal block f is 64, so in this case also 64 is the correct answer to be found. In this example, since a region having a significantly high luminance exists in the signal block g, the conventional correlation method is affected by the region. As a result, as shown in FIG.
The correlation function is not maximized at the correct position, and an erroneous motion vector is still detected.

On the other hand, in this embodiment, as shown in FIG. 10, the maximum value of C appears at the correct position 64, and it can be seen that the motion vector can be correctly detected even in this case.

In FIGS. 5 to 10, the luminance range of the signal is 0 to 255, and the luminance range when the signal is decomposed is
32 luminance ranges are set for each of eight gradations from luminance 0.

In this embodiment, the brightness range setting means 1
Stores 16 luminance ranges, and the input image is configured to be decomposed into 16 partial images in accordance with the stored luminance ranges. There may be one special case. In this case, the image decomposing means 2 extracts only one partial image having a luminance within this luminance range from the input image, and the correlation function calculating means 3 calculates a correlation function for this partial image.

The motion vector determining means 4 determines a motion vector from the maximum value of the correlation function. In this case, of course, the accuracy of the determined motion vector decreases, but on the other hand, there is an advantage that the processing load is reduced.

[Embodiment 2] Next, a second embodiment of the present invention will be described. FIG. 2 is a block diagram showing the configuration of the second embodiment of the present invention. Referring to FIG. 2, also in this embodiment, as in the case of the first embodiment, two previously blocked digital images, an image f (x, y) and an image g (x, y) are formed. Shall be entered.

Brightness range setting means 1 and image decomposition means 2
Is exactly the same as in the first embodiment.

In this embodiment, the phase correlation function calculating means 5
Calculates the phase correlation function for each image decomposed by the image decomposing means 2 in the following procedure.

That is, as in the first embodiment, first, the Fourier transforms Fj and Gj of the partial images f′j (x, y) and gj (x, y) are calculated, and then the complex of Fj is calculated. Normalized product of conjugate Fj * and Gj, Fj * xGj / | F
j * xGj | and, finally, the inverse Fourier transform of this product to calculate the phase correlation function dj (x, y)
Is calculated for each partial image.

The motion vector determining means 4 determines a motion vector by integrating the phase correlation functions dj (j = 1 to 16) calculated for each partial image. That is, the motion vector determining means 4 adds the respective phase correlation functions, and determines and outputs the final motion vector by detecting the position where the result of the addition takes the maximum value.

[Embodiment 3] Next, a third embodiment of the present invention will be described. FIG. 3 is a block diagram showing the configuration and processing of the third embodiment of the present invention. Also in this embodiment, as in the case of the first embodiment, it is assumed that two pre-blocked digital images, an image f (x, y) and an image g (x, y) are input. The processing of the luminance range setting means 1 and the image decomposing means 2 is the same as in the first embodiment.

In this embodiment, the motion vector calculating means 6
Calculates a motion vector for each image decomposed by the image decomposing means 2 in the following procedure.

That is, as in the first embodiment, the correlation function cj (x,
Calculate y).

Next, by detecting the position (xjmax, yjmax) at which it takes the maximum value from the correlation function cj (x, y), the motion vector vj obtained from the partial images fj, gj = (xjmax, yjmax) Is calculated for each partial image.

The motion vector determining means 4 determines and outputs the final motion vector v by calculating the average of these motion vectors.

Here, the vector average v is given by the following equation (8).

[0085]

(Equation 8)

Embodiment 4 Next, a fourth embodiment of the present invention will be described. FIG. 4 is a diagram showing the configuration and processing of the fourth embodiment of the present invention.

In the present invention, in general, a correlation function having a sharper peak can be obtained as the width of each luminance range is reduced when an input image is decomposed into partial images.
The accuracy of the determined motion vector is increased.

For example, FIG. 12 shows that in the first embodiment, the luminance range is divided by 0 to 1 gradation and a total of 25
The correlation function of each partial image is obtained using six luminance ranges, and the sum is calculated. The input signal is the same as in FIG. Compared to FIG. 7 in which the luminance range is set to 32 luminance ranges at intervals of 8 gradations and calculated, it can be seen that the peak is sharper.

However, when the width of each luminance range is set to be narrow as described above, the correlation function becomes sharp, but the operation becomes unstable with respect to the superposition of noise.

For example, the signal of FIG.
(See FIG. 11), when the width of each luminance range is set to 1, the sum of the correlation functions of the partial images becomes as shown in FIG. 13, and the correct position (position 64) Does not take the maximum value.

On the other hand, if the width of the luminance range is increased to some extent, it is possible to detect a motion vector stably even when noise is superimposed.

FIG. 14 shows the result of calculating the sum of the correlation functions of the partial images by setting the width of each luminance range to 16 for the same input as in FIG.

Referring to FIG. 14, the width of the correlation function is increased by increasing the width of the luminance range, but the maximum value is obtained at the correct position (position 64), and it can be seen that the motion vector can be correctly detected.

Therefore, in this embodiment, in view of the above points,
The brightness range setting means 1 includes a brightness range re-determining means 7,
The luminance range is set adaptively according to the input image.

In this embodiment, the luminance range redetermining means 7 redetermines the luminance range at a preset timing, for example, once every 5 seconds.

In other cases, a motion vector is determined by the same processing as in the first embodiment.

The redetermining process of the luminance range is performed as follows.

First, a plurality of brightness range candidates set in advance, for example, 128 brightness ranges for each brightness 2, 64 brightness ranges for each brightness 4, and 32 brightness ranges for each brightness 8
The luminance range is sequentially selected from five types of luminance range candidates, that is, 16 luminance ranges for each luminance 16, 16 luminance ranges for each luminance 16, and 8 luminance ranges for each luminance 32, and the selected luminance range is selected. , The input image is decomposed and the sum of the correlation functions of the partial images is calculated.

Further, the variance of the sum of the correlation functions of the partial images is calculated and stored.

After the variance of the sum of the correlation functions of the corresponding partial images is sequentially calculated for each luminance range candidate in this way, the luminance range with the minimum variance is selected, and this is determined as a new luminance range. .

[0101]

As described above, according to the present invention,
According to the plurality of luminance ranges set by the luminance range setting unit, the image decomposition unit decomposes the input image into a plurality of images, and then the motion vector calculation unit or the correlation function calculation unit performs The motion vector or the correlation function is calculated, and the motion vector determining means synthesizes the result to determine the final motion vector. This prevents the correlation function from being disturbed by the correlation, and as a result, the motion vector can be correctly calculated even in a case where there is an extremely high luminance region in the input image or in a region where the luminance changes gradually. This has the effect of making decisions.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration and a processing flow of a first embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration and a processing flow of a second embodiment of the present invention.

FIG. 3 is a block diagram showing a configuration and a processing flow of a third embodiment of the present invention.

FIG. 4 is a block diagram showing a configuration and a processing flow of a fourth embodiment of the present invention.

FIG. 5 is a diagram for explaining an embodiment of the present invention.

FIG. 6 is a diagram showing a result by a conventional correlation method as a comparative example.

FIG. 7 is a diagram for explaining an embodiment of the present invention.

FIG. 8 is a diagram for explaining an embodiment of the present invention.

FIG. 9 is a diagram showing a result of a conventional correlation method as a comparative example.

FIG. 10 is a diagram for explaining an embodiment of the present invention.

FIG. 11 is a diagram for explaining an embodiment of the present invention;
FIG. 5 is a diagram illustrating an influence of a setting method of a luminance range.

FIG. 12 is a diagram for explaining an embodiment of the present invention;
FIG. 5 is a diagram illustrating an influence of a setting method of a luminance range.

FIG. 13 is a diagram for explaining an embodiment of the present invention;
FIG. 5 is a diagram illustrating an influence of a setting method of a luminance range.

FIG. 14 is a diagram for explaining an embodiment of the present invention;
FIG. 5 is a diagram illustrating an influence of a setting method of a luminance range.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 brightness range setting means 2 image decomposition means 3 correlation function calculation means 4 motion vector determination means 5 phase correlation function calculation means 6 motion vector calculation means 7 brightness range re-determination means

Continuation of the front page (56) References JP-A-4-117089 (JP, A) JP-A-6-176151 (JP, A) JP-A-7-99661 (JP, A) JP-A-9-91453 (JP, A) , A) Kato et al., "Motion Compensated Prediction Coding Method Based on Equal Density Region", Journal of the Institute of Image Electronics Engineers of Japan, Vol. 23, No. 3, 1994, p. 157 −166 (58) Field surveyed (Int.Cl. ⁷ , DB name) G06T 7/20 H04N 7/32 JICST file (JOIS)

Claims (14)

(57) [Claims] 1. A brightness range setting means for setting a plurality of brightness ranges, and extracting pixels having brightness within the brightness range from an input image according to each of the brightness ranges, and decomposing the input image into a plurality of partial images. Image decomposition means, a correlation function calculation means for calculating a correlation function for each of the decomposed partial images, and a motion vector determination means for determining a motion vector by integrating the correlation functions for each of the partial images. A motion amount detection device characterized by the above-mentioned. 2. A luminance range setting means for setting a plurality of luminance ranges, a pixel having a luminance within the luminance range is extracted from an input image according to each of the luminance ranges, and the input image is decomposed into a plurality of partial images. An image decomposing means, a phase correlation function calculating means for calculating a phase correlation function for each of the decomposed partial images, and a motion vector determining means for determining a motion vector by integrating the phase correlation functions for each of the partial images. A motion amount detection device, comprising: 3. A brightness range setting means for setting a plurality of brightness ranges, and extracting pixels having brightness within the brightness range from an input image according to each of the brightness ranges, and decomposing the input image into a plurality of partial images. An image decomposing unit, a motion vector calculating unit that calculates a motion vector for each of the decomposed partial images, and a motion vector that integrates the motion vectors calculated for each of the partial images to determine a final motion vector A motion amount detection device, comprising: a determination unit. 4. The motion vector determining means determines a motion vector by detecting a pixel position where the sum of the correlation functions for each of the partial images added to all the partial images has a maximum value. The motion amount detecting device according to claim 1, wherein 5. The motion vector determining means determines a motion vector by detecting a position at which the sum of the sum of the phase correlation functions for each of the partial images for all the partial images has a maximum value. The motion amount detection device according to claim 2, wherein 6. The motion vector calculation means calculates a correlation function for each of the decomposed partial images and calculates a motion vector for each of the partial images by detecting a pixel position at which the correlation function has a maximum value. 4. The motion amount detecting apparatus according to claim 3, wherein the motion vector determining means determines a final motion vector by averaging the motion vectors for each of the partial images. 7. The motion amount detection device according to claim 1, wherein said luminance range setting means adaptively variably sets a luminance range according to said input image. apparatus. 8. The apparatus according to claim 1, wherein said brightness range setting means sets a predetermined brightness range.
The motion amount detection device according to any one of the above. 9. (a) extracting, for each of two input images, pixels having a luminance within each of a plurality of luminance ranges from the input image, thereby dividing the two input images; Respectively decomposing the image into partial images corresponding to the number of the luminance ranges; and (b) calculating a correlation function between the two input images for each of the decomposed partial images corresponding to the same luminance range. And (c) determining a motion vector by detecting a pixel position at which the sum of the correlation functions for each of the decomposed partial images for all of the partial images for each of the input images has a maximum value. Deciding a motion vector to be performed. 10. (a) For each of two input images,
Decomposing the two input images into partial images corresponding to the number of the luminance ranges, respectively, by extracting, from the input image, pixels having luminance within each luminance range for a plurality of luminance ranges; b) calculating a phase correlation function for each of the decomposed partial images corresponding to the same luminance range between the two input images; and (c) a phase correlation for each of the decomposed partial images. Determining a motion vector by detecting a pixel position at which the sum of functions added to all of the partial images per input image has a maximum value, thereby determining a motion vector. 11. (a) For each of two input images,
Extracting, from the input image, pixels having a luminance within the luminance range according to each luminance range of the plurality of luminance ranges, and decomposing the two input images into partial images corresponding to the number of the luminance ranges. And (b) calculating a motion vector for each of the decomposed partial images between the two input images; and (c) integrating the motion vectors calculated for each of the partial images to obtain a final motion. Determining a vector; and a motion amount detecting method. 12. In the step (b), a correlation function for each of the decomposed partial images corresponding to the same luminance range is obtained between the two input images, and the correlation function takes a maximum value. A motion vector is calculated for each partial image by detecting a position, and in step (c), a final motion vector is determined by averaging the motion vectors for each of the partial images. The method of detecting a motion amount according to claim 11, wherein The method according to claim 13, wherein the luminance range in response to the input image, adaptively variably set, the motion amount detection method according to any one of claims 9 to 1 1, characterized in that. 14. The method of claim 13, wherein the luminance range, the luminance within a predetermined range, the motion amount detection method according to any one of claims 9 to 1 1, characterized in that.