JPH07336691A - Motion vector detecting device - Google Patents

Abstract

PURPOSE:To reduce image distortion due to the erroneous detection of a deflection vector. CONSTITUTION:A motion vector detecting part 13 detects two motion vectors V1, V2 by repeating gradient method calculation twice as considering an initial deflection vector V0 selected from among the detected motion vectors to be a starting point. A difference value comparing part 14 compares the sizes of an inter-field difference value DED0 based on the initial deflection vector V0 and the interfield difference value DFW based on the motion vector V1 obtained by the gradient method calculation of a first time. A motion vector selecting part 15 selects one of the initial deflection vector V0 and the motion vector V2 obtained by the gradient method calculation of a second time as the true motion vector on the basis of the result of this comparison.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention divides a television signal into blocks each having a predetermined number of pixels and a predetermined number of lines, and for each block, a moving image is produced between signals separated by a predetermined interval in the time axis direction. The present invention relates to a motion vector detection device that detects a motion vector of an image.

[0002]

2. Description of the Related Art Generally, when performing interframe coding processing for high efficiency coding or field interpolation processing for field number conversion, a motion vector indicating the direction and magnitude of motion of a moving image is used. To be

As a method for detecting this motion vector, a method using a pattern matching method (Japanese Patent Laid-Open No. 55-162)
683, JP-A-55-162684) and a method using an iterative gradient method (JP-A-60-158786).

The method using the pattern matching method detects a motion vector by utilizing the similarity of signal patterns between frames. In the method using the iterative gradient method, the amount of motion is estimated by utilizing the physical correspondence between the gradient of the intra-frame signal and the difference value of the inter-frame signal.

By the way, in the method using the iterative gradient method,
In order to improve the detection accuracy of the motion vector, a method using an initial displacement vector has been considered (Japanese Patent Laid-Open No. 62-
206980).

In this method, a predetermined vector is selected as an initial displacement vector from a plurality of motion vectors that have already been detected, and this initial displacement vector is used as a starting point.
The true motion vector is detected by repeating the gradient method calculation a plurality of times.

[0007]

However, in the conventional motion vector detecting device that detects a motion vector using the initial deviation vector, if the deviation vector detected by the gradient method calculation is erroneously detected, it is not possible. There is a problem that an image is distorted by performing field interpolation processing or the like with an appropriate motion vector. In particular, when the displacement vector is detected by an arithmetic expression as in the iterative gradient method, the displacement vector is likely to be erroneously detected depending on the image, and thus the above-mentioned problem is likely to occur.

[0008]

The motion vector detecting device of the present invention comprises a motion vector detecting means, a difference value comparing means, and a motion vector selecting means.

Here, the motion vector detecting means repeats a predetermined displacement vector detection process a plurality of times starting from an initial displacement vector selected from among a plurality of motion vectors that have already been detected, to thereby obtain a plurality of motion vectors. Detect a motion vector.

The difference value comparing means compares the difference value in the time axis direction of the television signal motion-compensated with at least one motion vector of the plurality of motion vectors and the television signal motion-compensated with the initial deviation vector. The magnitude of the difference value in the time axis direction is compared.

Based on the comparison result, the motion vector selection means selects one of the plurality of motion vectors detected by the final displacement vector detection processing and the initial displacement vector as a true motion vector. Select as a vector.

[0012]

In the above arrangement, the predetermined displacement vector detection process is repeated twice, for example, starting from the initial displacement vector, and two motion vectors are detected.

Next, for example, the difference value in the time axis direction of the television signal motion-compensated by the motion vector detected by the first displacement vector detection process and the television signal motion-compensated by the initial displacement vector. The magnitudes of the difference values in the time axis direction are compared.

Next, based on this comparison result, for example,
Either the motion vector detected by the final displacement vector detection process or the initial displacement vector is detected as the true motion vector.

That is, when the difference value based on the motion vector is larger than the difference value based on the initial displacement vector, the initial displacement vector is selected as the true motion vector, and when the difference value based on the initial displacement vector is small, the motion vector is selected as the true motion vector. It

[0016]

Embodiments of the present invention will now be described in detail with reference to the drawings. FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention.

The illustrated motion vector detecting device comprises input terminals 11 and 12, a motion vector detecting section 13, a difference value comparing section 14, a motion vector selecting section 15, and an output terminal 16.

In such a structure, the motion vector detecting device divides the television signal into a plurality of blocks each consisting of m pixels in the horizontal direction and n lines in the vertical direction (m and n are integers) for a total of m × n pixels. , For each block, for example,
A motion vector of a moving image is detected between signals separated by one field in the time axis direction.

That is, in FIG. 1, the input terminal 11 is supplied with a television signal A of a field (hereinafter referred to as "reference field") which is a reference when detecting a motion vector. On the other hand, the input terminal 12 is supplied with the television signal B of a field for detecting a motion vector, which is located one field before or after the reference field (hereinafter referred to as "detection field").

The motion vector detection unit 13 includes an initial displacement vector selection circuit 131, two gradient method operation circuits 132,
And an adder circuit 133, 135.

In such a configuration, the motion vector detecting section 13 selects one of a plurality of motion vectors already detected for each block as the initial displacement vector V0, which will be described in detail later. The gradient method calculation is repeated twice with the initial displacement vector V0 as a starting point to detect the two motion vectors V1 and V2.

The difference value comparison unit 14 includes two vector correction circuits 141 and 142 and two subtraction circuits 143 and 144.
2, absolute value calculation circuits 145, 146, two cumulative addition circuits 147, 148, and a comparison circuit 149.

In such a configuration, the difference value comparison unit 1
4, a field of the television signal motion-corrected by the motion vector V1 detected by the first gradient method operation of the two motion vectors V1, V2 detected by the motion vector detection unit 13 will be described later in detail. The magnitude of the inter-field difference value DFD0 of the television signal motion-corrected by the initial difference vector V0 is compared with the inter-field difference value DFDα.

The motion vector selection unit 15 selects two of the two motion vectors V1 and V2 detected by the motion vector detection unit 13 based on the comparison result of the difference value comparison unit 14.
Either the motion vector V2 detected by the second gradient method calculation or the initial displacement vector V0 is set as the true motion vector V
It is supposed to be selected as.

The operation of the above configuration will be described. First, the motion vector detecting operation of the motion vector detecting unit 13 will be described. The field signals A and B supplied to the input terminals 11 and 12 are supplied to the initial displacement vector detection circuit 131 and the gradient method calculation circuits 132 and 134.

The initial displacement vector selection circuit 131 selects, for each block, one of a plurality of motion vectors that have already been detected as the initial displacement vector V0.

Here, as the motion vector already detected, for example, the motion vectors of a plurality of blocks located in the vicinity of the block for which the motion vector V is currently detected, the average motion vector of the reference field, and further, , Acceleration motion vectors, etc. are used. Further, as a method of selecting any one of the plurality of already detected motion vectors as the initial displacement vector V0, for example, a pattern matching method is used.

The initial displacement vector V0 selected by the initial displacement vector selection circuit 131 is the gradient method operation circuit 13
2 is supplied. As a result, the first gradient method calculation is performed starting from the initial displacement vector V0, and the displacement vector α is obtained. This deviation vector α is calculated by the addition circuit 13
3 and is added to the initial displacement vector V0. As a result, the motion vector V1 (= V0 + α) is obtained.

This motion vector V1 is supplied to the gradient method calculation circuit 134. As a result, the second gradient method calculation is performed with the motion vector V1 as the starting point, and the displacement vector β is obtained. This deviation vector β is added to the addition circuit 13
5 and is added to the motion vector V1. As a result, the motion vector V2 (= V1 + β = V0 + α + β)
Is obtained.

FIG. 2 is a vector diagram showing the above situation. As shown, the motion vector V1 is obtained by combining the initial displacement vector V0 and the displacement vector α, and the motion vector V2 is obtained by combining the motion vector V1 and the displacement vector β.

The above is the motion vector detecting operation of the motion vector detecting unit 13. Next, the difference value comparison operation of the difference value comparison unit 14 will be described.

The detection field signal B supplied to the input terminal 12 is supplied to the vector correction circuit 141 and subjected to motion correction using the initial displacement vector V0. As a result, the detection field signal C0 in which the coordinates of the moving image included in the detection field signal B are shifted by the initial displacement vector V0 is obtained.

Further, the detection field signal B supplied to the input terminal 12 is further supplied to the vector correction circuit 142 and subjected to motion correction using the motion vector V1. As a result, the detection field signal Cα obtained by shifting the coordinates of the moving image included in the detection field signal B by the motion vector V1 is obtained.

The detection field signal C0 output from the vector correction circuit 141 is supplied to the subtraction circuit 143,
It is subtracted from the reference field signal A supplied to the input terminal 11 for each pixel. As a result, for each pixel,
An inter-field difference value D0 indicating the difference between the detection field signal C0 and the reference field signal A is obtained.

Similarly, the detection field signal Cα output from the vector correction circuit 142 is supplied to the subtraction circuit 143 and subtracted from the reference field signal A supplied to the input terminal 11 for each pixel. As a result, the inter-field difference value Dα between the detection field signal Cα and the reference field signal A is obtained for each pixel.

The signals indicating the inter-field difference values D0 and Dα are supplied to the corresponding absolute value circuits 145 and 146, and their absolute values E0 and Eα are taken. Each absolute value E
The signals indicating 0 and Eα are supplied to the corresponding cumulative addition circuits 147 and 148, respectively, and cumulatively added within the range of the motion vector detection block.

As a result, inter-field difference values DFD0 and DFDα in block units are obtained. The signals indicating the inter-field difference values DFD0 and DFDα are supplied to the comparison circuit 14
9 and compared in size.

The above is the difference value comparing operation of the difference value comparing section 14. Next, the motion vector selecting operation of the motion vector selecting unit 15 will be described.

The motion vector selection unit 15 selects either the initial displacement vector V0 or the motion vector V2 as the true motion vector V based on the comparison result G of the inter-field difference values.

In this case, if the inter-field difference value DFDα based on the motion vector V1 is larger than the inter-field difference value DFD0 based on the initial displacement vector, the initial displacement vector V0 is selected as the true motion vector V. On the other hand, when the inter-field difference value DFDα is smaller than the inter-field difference value DFD0, the motion vector V2 is selected as the true motion vector V.

Accordingly, when it is highly likely that the displacement vectors α and β are erroneously detected, the true displacement vector V is not the motion vector V2 but the initial displacement vector V.
Since 0 is selected, it is possible to reduce the occurrence of image distortion due to erroneous detection of the displacement vectors α and β.

According to this embodiment described in detail above, the following effects can be obtained. (1) Interfield difference value DF based on motion vector V1
Dα and the size of the inter-field difference value DFD0 based on the initial displacement vector V0 are compared. If the former is larger, the initial displacement vector V0 is selected as the true motion vector V, and if smaller, the motion vector V2 is the true motion vector. Since it is selected as the vector V, it is possible to reduce the occurrence of image distortion due to erroneous detection of the deviation vectors α and β.

(2) Since the field difference value DFDα by the motion vector V1 detected by the first gradient method calculation is used as the inter-field difference value by the motion vector, it is detected by the second gradient method calculation. Compared with the case where the inter-field difference value based on the motion vector V2 is used, the time until the comparison result is obtained can be shortened.

As a result, when the field signals A and B are read from the image memory (not shown), the number of pixels to be read at one time can be reduced, so that the hardware amount of the read circuit can be reduced.

Although one embodiment of the present invention has been described in detail above, the present invention is not limited to the above embodiment.

(1) For example, in the above embodiment, the case where the inter-field difference value DFDα by the motion vector V1 obtained by the first gradient method operation is used as the inter-field difference value by the motion vector has been described. However, the present invention may use the inter-field difference value based on the motion vector V2 obtained by the second gradient method calculation.

According to such a configuration, the displacement vector α
Even when the deviation vector β is not erroneously detected, the motion vector V2 detected by the iterative gradient method can be prevented from being selected as the true motion vector V.

(2) In the above embodiment, the case where either the initial displacement vector V0 or the motion vector V2 obtained by the second gradient method operation is selected as the true motion vector V has been described. .

However, according to the present invention, the initial displacement vector V
The inter-field difference value by 0, the inter-field difference value by the motion vector V1 obtained by the first gradient method operation, and the inter-field difference value by the motion vector V2 obtained by the second gradient method operation are compared. Alternatively, any one of these three vectors V0, V1, V may be selected as the true motion vector V.

According to this structure, since the selection target of the true motion vector V can be increased, the accuracy of the true motion vector V can be improved.

(3) In the above embodiment, the gradient method calculation is repeated twice to obtain two motion vectors V1,
The case of detecting V2 has been described. However, this invention
It is also possible to detect three or more motion vectors by repeating the gradient method calculation three times or more.

(4) In the above embodiment, the case where a plurality of motion vectors are detected by repeating the gradient method calculation a plurality of times has been described. However, the present invention detects one motion vector V1 by executing the gradient method operation only once, and selects either the motion vector V1 or the initial displacement vector V0 as the true motion vector V. You may do it.

(5) In the above embodiment, the gradient method calculation process is used as the deviation vector detection process. However, the present invention may use processing other than the gradient method calculation processing, for example, processing using the pattern matching method.

(6) In the above embodiment, the case where the present invention is applied to a motion vector detecting device for detecting a motion vector between signals separated by one field in the time axis direction has been described. However, the present invention can also be applied to a motion vector detection device that detects a motion vector between signals separated by two or more fields.

(7) In addition to this, it is needless to say that the present invention can be variously modified without departing from the scope of the invention.

[0056]

As described in detail above, according to the present invention, the magnitude of the difference value in the time axis direction by the motion vector and the difference value in the time axis direction by the initial displacement vector are compared, and the result of this comparison is compared. Since the true motion vector is selected based on the above, it is possible to reduce the occurrence of image distortion due to erroneous detection of the displacement vector.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention.

FIG. 2 is a vector diagram for explaining the operation of a motion vector detection unit of one embodiment.

[Explanation of symbols]

 11, 12 ... Input terminal 13 ... Motion vector detection unit 14 ... Difference value comparison unit 15 ... Motion vector selection unit 16 ... Output terminal 131 ... Initial deviation vector selection unit 132, 134 ... Gradient method operation circuit 133, 135 ... Addition circuit 141, 142 ... Vector correction circuit 143, 144 ... Subtraction circuit 145, 146 ... Absolute value calculation circuit 147, 148 ... Cumulative addition circuit 149 ... Comparison circuit

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Yuji Nojiri 2-2-1, Jinnan, Shibuya-ku, Tokyo Within the Japan Broadcasting Corporation (72) Hiroshi Hirabayashi 2-2-1, Jinnan, Shibuya-ku, Tokyo Within the Sending Association (72) Gen Sonehara 2-2-1, Jinnan, Shibuya-ku, Tokyo Inside the Sending Association of Japan

Claims (3)

[Claims] 1. A television signal is divided into blocks each having a predetermined number of pixels and a predetermined number of lines, and a motion vector of a moving image is detected for each block between signals separated by a predetermined interval in the time axis direction. In the motion vector detection device, for each block, a predetermined displacement vector detection process is repeated a plurality of times starting from an initial displacement vector selected from a plurality of motion vectors that have already been detected. A motion vector detecting means for detecting a motion vector, a difference value in the time axis direction of the television signal motion-corrected by at least one motion vector among the plurality of motion vectors detected by the motion vector detecting means, and the initial deviation. Difference value that compares the magnitude of the television signal motion-corrected by the position vector with the difference value in the time axis direction Comparing means, and based on the comparison result of the difference value comparing means, the motion vector detected by at least the last displacement vector detecting process among the plurality of motion vectors detected by the motion vector detecting means and the initial displacement. A motion vector detecting device comprising: a motion vector selecting means for selecting one of the vectors as a true motion vector. 2. The time of the television signal motion-compensated by the difference value comparison means by the motion vector detected by the first displacement vector detection process among the plurality of motion vectors detected by the motion vector detection means. It is configured to compare the magnitude of the difference value in the axial direction with the difference value in the time axis direction of the television signal motion-corrected by the initial deviation vector, and the motion vector selection means is the difference value comparison means. Based on the result of the comparison, one of the motion vector detected by the final displacement vector detection process among the plurality of motion vectors detected by the motion vector detection means and the initial displacement vector is a true motion vector. The motion vector detection device according to claim 1, wherein the motion vector detection device is configured to be selected as. 3. A television signal is divided into blocks each having a predetermined number of pixels and a predetermined number of lines, and a motion vector of a moving image is detected for each block between signals separated by a predetermined interval in the time axis direction. In the motion vector detecting device, for each block, a predetermined displacement vector detection process is executed with an initial displacement vector selected from a plurality of already detected motion vectors as a starting point, whereby one motion is detected. Motion vector detecting means for detecting a vector, and a difference value in the time axis direction of the television signal motion-compensated by the motion vector detected by the motion vector detecting means and the television signal motion-compensated by the initial deviation vector. Based on the comparison result of the difference value comparison means and the difference value comparison means for comparing the magnitude with the difference value in the time axis direction of Furthermore, the motion vector detection device is provided with a motion vector selection unit that selects one of the motion vector detected by the motion vector detection unit and the initial displacement vector as a true motion vector.