JP3397486B2 - MOTION VECTOR MOTION VECTOR DETECTION METHOD AND MOTION VECTOR DETECTION CIRCUIT - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a motion vector detecting method and a motion vector detecting circuit for detecting a motion vector of a moving object from a moving image signal, and more particularly to a motion vector detecting circuit.
For example, the present invention relates to a motion vector detection method and a motion vector detection circuit of a moving image signal suitable for use in a television system conversion device for performing signal conversion between different television systems or a high efficiency coding device in image transmission.

[0002]

2. Description of the Related Art A technique for detecting a moving area in a moving image as a motion vector and correcting the motion using the motion vector is a technique for improving interframe coding efficiency in high-efficiency coding during image transmission. Or, it is effectively used for reducing motion discontinuity due to conversion of the number of fields during conversion of the television system. For example, as a television system conversion device, Japanese Patent Laid-Open Nos. 01-309597 and 03-280681, the Journal of the Television Society of 1991 (Vol. 45, No. 12, page 1534 to page 1543), Or 1989
Proceedings (20-
The ones described in 5) are known.

Conventionally, a motion vector detecting method applied to these devices is arranged such that each field of a digitized television signal has a predetermined size of (mxn) pixels including m pixels in the horizontal direction and n lines in the vertical direction. Each block is divided into blocks, and the amount of motion is calculated by iterative gradient calculation based on the gradient of the signal in the screen and the physical correspondence of the signal difference value between the corresponding screens. The so-called iterative gradient method
The motion vector was detected by an arithmetic method called ent method). In this case, as the initial value when estimating the motion amount, the optimum one of the motion vectors already detected before the detected block is selected as the initial displacement vector and set as the true motion vector of the detected block. By decreasing the number of gradient calculations by starting the calculation from a close value,
For example, a circuit that detects a true motion vector by performing two calculations has been realized.

In this case, as the candidates for the initial displacement vector, the motion vector (adjacent vector) of the detected block adjacent to the detected block from the same field, or the block corresponding to the detected block in the temporally previous field is used. An average vector obtained by averaging the motion vectors of a plurality of blocks including the acceleration vector, an acceleration vector represented by the difference between the average vector of the previous field and twice the average vector of the previous field, and the like are selected as candidates. It was To find the optimum initial displacement vector among these, the coordinates of the corresponding block in the previous field are displaced by the value of each initial displacement vector, and the difference between the displaced field signal and the current field signal is calculated. Then, the one in which the value obtained by accumulating the absolute value for the number of pixels of the block is the minimum is selected as the optimum initial displacement vector.

However, in the above apparatus, many candidate vectors, for example, six types of motion vectors must be accumulated as candidates for the initial displacement vector, and many arithmetic results must be accumulated when selecting them, so that a large number of memories are required. Was required, and the circuit scale was increasing. Therefore, for example, in the motion vector detection method described in Japanese Patent Laid-Open No. 04-78286, the initial displacement that reduces the circuit scale by reducing the candidates for the initial displacement vector by utilizing the characteristics of the television signal is proposed. A motion vector detection method using a vector has been proposed. In this method, focusing on the fact that a moving image represented by a television signal has a constant speed and a large number of parts moving in a single direction, at least one already-detected motion vector having a large difference from the average vector is selected, This and the average vector are selected as candidates for the initial displacement vector, and the optimum one is selected as the initial displacement vector. As a result, the size of the circuit has been reduced by reducing the memory for accumulating candidates for the initial displacement vector and the memory for accumulating the calculation result thereof to about two types.

[0006]

However, in the above-mentioned conventional technique, although the circuit is downsized, since the initial deviation vector based on the average vector is adopted, the entire image area is reduced. When an image signal including a moving object having a relatively small area is input, there is a problem that the motion vector of the moving object may not be accurately detected. For example, if the moving object is smaller than the detected block, the value of the motion vector of the block adjacent to the detected block will be "0", and the value of the average vector will also be close to "0". When these are selected as the initial displacement vector, the initial displacement vector for a block including a moving object having a small area is almost selected as a motion "0" vector. Therefore, starting from the initial displacement vector where the vector value is “0”, for example, the value of the true motion vector cannot be reached in two gradient calculations, and most of the area is the static area of the motion vector “0”. There is a drawback that the animal body may not be detected because it is included in the part.

The present invention solves the above problems and provides a motion vector detecting method and a motion vector detecting circuit for a moving image signal, which is capable of downsizing the apparatus and accurately detecting a moving object having a small area. The purpose is to

[0008]

In order to solve the above-mentioned problems, a method for detecting a motion vector of a moving image signal according to the present invention is applied to each field when detecting a motion vector of a moving object in a digitized image signal. Is subdivided into blocks containing multiple pixels and lines, and an initial displacement vector that is the initial value for estimating the amount of motion is set for each block, and each detected target is detected based on this initial displacement vector. In the motion vector detection method of the moving image signal for sequentially detecting the motion vector of the block,
The first step of accumulating the motion vector of the detected block for at least one field, and the motion vector accumulated in the first step are read out together with the motion vectors of the surrounding blocks, and these are averaged for each block. First
In the second step of determining the average vector of the above, the third step of determining whether or not the value of the first average vector obtained in the second step is “0”, and the second step A fourth step of determining whether or not the value of the first average vector obtained as described above exceeds a predetermined value preset according to the image, and the predetermined value determined in the fourth step. The first average vector that exceeds 1 is accumulated over one field, and these are averaged over the entire field.
Of the first average vector as a candidate for the initial displacement vector of the detected block when it is determined that the value of the first average vector is not "0" in the fourth step A sixth step of selecting a vector and selecting the second average vector obtained in the fifth step as a candidate for the initial deviation vector when it is determined that the first average vector is “0” And a motion vector of a detected block adjacent to the candidate block of the initial displacement vector selected in the sixth step and a detected block, and a value of the value closest to the value of the motion vector of the detected block. The seventh step of selecting the motion vector as the initial displacement vector, the initial displacement vector selected in the seventh step, the image signal of the current field and the image signal of the previous field are received and based on these. And having an eighth step of detecting a true motion vector of the detected block Te.

In this case, in the seventh step of the present invention, the image signal of the block of the previous field corresponding to the detected block is coordinate-converted for each pixel at each value of the candidates of the initial displacement vector. The step, the step of obtaining the difference value between the coordinate-transformed signal and the image signal of the detected block for each pixel, and the absolute value of the difference value accumulated for each pixel are compared, and the accumulated value is the minimum. And the step of selecting a motion vector that becomes as the initial displacement vector.

Further, in this case, the seventh step compares the average vector selected in the sixth step with the motion vector of the detected block adjacent to the detected block, and determines the difference between the average vector and the motion vector. It is preferable to include a step of selecting at least one motion vector of a large already detected block and selecting the motion vector of this already detected block and the average vector of the sixth step as candidates for the initial displacement vector.

Further, in the eighth step, the image signals of the previous field and the current field are received, and the signal obtained by displacing the image signal of the previous field by the initial displacement vector selected in the seventh step is used as a starting point. The true motion vector of the detected block may be detected by repeatedly calculating the amount of motion deviation with respect to the image signal of the current field based on the gradient amount and the difference amount of each signal.

On the other hand, the motion vector detection circuit according to the present invention, when detecting the motion vector of the moving object in the digitized image signal, subdivides each field into blocks including a plurality of pixels and lines, In the motion vector detection circuit that sets the initial displacement vector that is the initial value when estimating the motion amount for each block of, and detects the motion vector based on this initial displacement vector, at least the motion vector of the already detected block is 1
Storage means for accumulating fields, and first average vector calculation means for obtaining the first average vector for each block by reading out the motion vectors accumulated in the storage means together with the motion vectors of the surrounding blocks and averaging them , A first determination unit that determines whether or not the values of the first average vector calculated by the first average vector calculation unit are “0”, and a first determination unit that is calculated by the first calculation unit.
A second determination of whether or not the value of the average vector of the vector exceeds a predetermined value set in advance according to the image.
And the first average vector exceeding the predetermined value determined by the second determining means are accumulated over one field to obtain a second average vector obtained by averaging them over the entire field. When the second average vector calculation means and the first determination means determine that the value of the first average vector is not “0”, the first average vector as a candidate for the initial displacement vector of the detected block And selecting the first average vector as “0”, the second average vector obtained by the second average vector calculating means is selected as a candidate for the initial deviation vector. Of the initial deviation vector selected by the first selecting means and the motion vector of the detected block adjacent to the detected block, and the motion of the detected block among them is received. Second selection means for selecting a motion vector having a value closest to the value of the vector as the initial displacement vector, the initial displacement vector selected by the second selection means, the image signal of the current field, and the image signal of the previous field. In response to this, there is provided a vector detecting means for detecting the true motion vector of the detected block based on these.

In this case, the second selecting means of the present invention coordinate-converts the image signal of the block of the previous field corresponding to the detected block for each pixel at each value of the candidates of the initial displacement vector. A conversion means for calculating the difference value between the coordinate-converted signal and the image signal of the detected block for each pixel, accumulating means for accumulating the absolute value of the difference value for the number of pixels of the block, It is preferable to include a comparing unit that compares the outputs of the accumulating unit and a selecting unit that receives the result of the comparing unit and selects the motion vector having the smallest accumulated value as the initial displacement vector.

The second selecting means compares the average vector selected by the first selecting means with the motion vector of an already detected block adjacent to the detected block, and the second vector has a large difference from the average vector. There is a vector comparison means for selecting at least one motion vector of the detection block, and either one of the motion vector of the already detected block selected by this vector comparison means and the average vector selected by the first selection means. The optimum one should be selected as the initial deviation vector.

Further, the vector detecting means receives the image signals of the previous field and the current field, and starts a signal obtained by displacing the image signal of the previous field by the initial displacement vector selected by the second selecting means. Therefore, it is preferable to have a plurality of gradient calculating means for calculating the movement deviation amount for the image signal of the current field based on the gradient amount and the difference amount of each signal.

Further, the motion vector detection circuit of the present invention includes a vector evaluation means for evaluating whether or not the motion vector detected by the vector detection means is detected within a predetermined error range. It is characterized in that only the motion vector which becomes good is stored in the output and storage means.

[0017]

According to the motion vector detecting method and the motion vector detecting circuit of the present invention, the motion vector of the detected block of the previous field corresponding to the detected block of the current field and the motion vectors of the peripheral blocks are averaged. The determined first average vector is sequentially obtained, and it is determined whether or not the first average vector is "0". If the first average vector is not "0", the first average vector is determined. Is selected as a candidate for the initial displacement vector. On the other hand, it is determined whether or not the first average vector exceeds a predetermined value set according to the image, and the first average vector determined to exceed the predetermined value is accumulated over one field. Then, the accumulated first average vector is averaged in the field to calculate the second average vector.
Then, the portion where the value of the first average vector is determined to be “0” is replaced with the second average vector as a candidate for the initial displacement vector, and this is selected. As a result, an effective value corresponding to the image is set in the selected portion where the average vector becomes “0”, and further, the selected average vector and the motion vector in the already detected block of the current field are set. The optimum one is selected as the initial displacement vector. In particular, when the area of the moving body is small, in a block including the moving body or a block near the boundary between the static region and the moving region, the second motion vector close to the motion vector of the image of that part is used.
Is selected, and the true motion vector of the detected block is effectively detected from the corresponding block in the previous field based on this initial displacement vector, for example, by the iterative gradient method.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a motion vector detecting method and motion vector detecting circuit for an image signal according to the present invention will be described in detail with reference to the accompanying drawings. 3 to 5 show a basic circuit and a principle diagram applied to the motion vector detecting method of the image signal and the motion vector detecting circuit in the present invention. First, in order to facilitate understanding of the present embodiment,
The principle of the basic motion vector detection method will be described with reference to these drawings.

The current field signal 1 for which the motion vector is to be obtained and the corresponding previous field signal 2
(See FIG. 4) is input to this circuit via the two-dimensional low-pass filters 49, 49 and supplied to the initial displacement vector selection circuit 42 and the two gradient method operation circuits 44, 45, respectively.
In these circuits 42, 44 and 45, the field signals 1 and 2 are subdivided into blocks each having m pixels in the horizontal direction and n pixels in the vertical direction, and the respective processes are performed.

The two-dimensional low-pass filter 49 removes noise from the field signal and reduces high frequency components. The initial displacement vector selection circuit 42 reads the motion vector detected up to the previous time from the motion vector memory circuit 41, and selects the motion vector closest to the motion vector of the detected block of the current field signal from the motion vector memory circuit 41 as the initial displacement vector. It is selected as the position vector Vo (see FIG. 5) and supplied to the first gradient method computing circuit 44.

The first gradient method operation circuit 44 estimates the amount of movement of the current field from the coordinates where the previous field signal 2 is displaced by the initial displacement vector Vo as the starting point, and the first movement displacement amount V1 (See FIG. 5). This output is added to the initial displacement vector Vo by the adder circuit 46 and supplied to the second gradient method operation circuit 45. Similarly, the second gradient method operation circuit 45
Calculates and outputs the second deviation V2 as in the circuit 44.
This output is added by the adder circuit 47 to the output of the adder circuit 46, and the value is detected as the motion vector V (= Vo + V1 + V2) of the detected block in the current field 1 (see FIG. 5) and output.
48 and the motion vector memory 41, respectively.
As a result, the amount of movement of the object located at the coordinates (m1, n1) in the previous field to the coordinates (m1 + α0, n1 + β0) in the current field is detected as the vector V.

At this time, the gradient method operation circuits 44 and 45 obtain respective deviations by the following operation expressions (1) to (4), for example. Vx = ΣSGN ΔX ・ DFD / Σ | ΔX | ・ ・ ・ ・ (1) Vy = ΣSGN ΔY ・ DFD / Σ | ΔY | ・ ・ ・ ・ (2) ΔX = (A _{n + 1, m} -A _{n-1 , m} ) / 2 (3) ΔY = (A _{n, m + 1} -A _{n, m-1} ) / 2 (4) where Vx is the x-direction component of the motion vector V, Vy is the y-direction component of the motion vector V, A _{n, m} is the n-pixel, m-line coordinate signal, ΔX is the gradient in the X direction of the image, ΔY is the gradient in the Y direction of the image, and DFD is the interfield difference value. Show. Also SG
N ΔX, SGN ΔY indicates a code including “0” in each of ΔX and ΔY.

That is, in the gradient method operation circuits 44 and 45, the movement deviation is generated based on the gradient values in the X direction and the Y direction of the image signals of the current field and the previous field and the difference values of the signals from the initial displacement vector Vo as a starting point. The unit amount is obtained, and the calculation is repeated to estimate the true amount of movement by the above calculation. The resulting motion amount, that is, the deviation amounts V1 and V2 are output from the respective circuits 44 and 45. Therefore, an accurate motion vector V can be obtained by supplying the initial displacement vector Vo to the gradient method operation circuits 44 and 45 as an accurate value.

FIG. 1 shows an embodiment of a motion vector detection circuit to which the motion vector detection method for moving picture signals according to the present invention is applied. The motion vector detection circuit of the present embodiment is based on the above-mentioned circuit as a basic configuration, and in particular, the initial displacement vector selection circuit 42 selects and outputs an accurate initial displacement vector without increasing candidates for the initial displacement vector. The main feature is that it does. In this embodiment, the same parts as those in FIG. 3 are designated by the same reference numerals.

In the motion vector detecting circuit of this embodiment, the current field signal 1 and the previous field signal 2 supplied via the low pass filters 49, 49 are the same as above.
Is input via the input terminals 10 and 12, and is supplied to the initial displacement vector selection unit 42, the vector detection unit 50, and the vector evaluation unit 60. The vector detection unit 50 includes the gradient calculation circuit 4 described above.
An arithmetic circuit including 4,45, detects a motion vector based on the initial displacement vector C from the initial displacement vector selection circuit 42, and supplies the output D as a result to the vector evaluation unit 60. The vector evaluation unit 60 evaluates whether or not the value of the motion vector detected by the vector detection unit 50 is detected within a predetermined error range, and outputs only the motion vector E whose evaluation result is good. This is a vector determination circuit for outputting from the terminal 14 and storing it in the vector memory circuit 41.

The vector evaluation unit 60 of the present embodiment, for example, has two field memories 151a and 151a as shown in FIG.
b, the subtraction circuit 152, the absolute value circuit 154, and the accumulation circuit 15
5, a comparison circuit 156, and a switching circuit 157. memory
151a is a field memory for sequentially accumulating the current field signal 1 for each field. The memory 151a is a field memory that stores the previous field signal 2 and outputs a signal obtained by coordinate-converting the signal of each block with the motion vector D. The subtraction circuit 152 is the current field memory.
It is an arithmetic circuit that subtracts the image signal of the previous field memory 151b from the image signal of 151a. The absolute value circuit 154 is an arithmetic circuit that converts the difference signal of each block into an absolute value for each pixel and outputs the absolute value. The accumulation circuit 155 is an addition circuit that accumulates the output from the absolute value circuit 154 by the number of pixels in each block. The comparison circuit 156 is a determination circuit that compares the output of the result from the accumulation circuit 155 with a threshold value “A” to determine whether the motion vector is good or bad. The switching circuit 157 is a comparison circuit 156.
Is an output circuit that outputs the motion vector D as a true motion vector E when the comparison result from 1 is good.

Returning to FIG. 1, the motion vector E from the vector evaluation unit 60 is stored in the vector memory for each block.
Sequentially accumulated in 41. The vector memory 41 is formed of a storage circuit such as a FIFO memory, which stores a motion vector for at least one field. The initial displacement vector selection unit 42 reads the motion vector of the already detected block from the vector memory 41 and selects the optimum initial displacement vector for detecting the detected block. In particular, the initial deviation vector selection unit 42 of the present embodiment, as shown in this figure, includes a first average vector calculation unit 102, a first average vector determination unit 104, and a second average vector determination unit 106. ,
Second average vector calculation unit 108, first initial displacement vector selection unit 110, second initial displacement vector selection unit 112
And have.

Specifically, the first average vector calculation unit 102
Is an arithmetic circuit for reading the motion vector of the detected block of the previous field corresponding to the position of the detected block and the motion vector F of its periphery from the vector memory 41 and obtaining the average value thereof for each block. For example,
In the case of the present embodiment, the value of the motion vector of the corresponding block is set to "C
1 ”, the value of the motion vector of the surrounding blocks is“ C2 ”to“ C
9 ", these are added and divided by the number 9 of the blocks to obtain an average vector H (hereinafter, referred to as a first average vector H).

The first average vector determination unit 104 is a determination circuit for determining whether or not the value of the first average vector H is "0", and the determination result is the first initial deviation vector selection. The selection signal K is output to the section 110. For example, the selection signal K is a signal which is turned on when the value of the first average vector is "0" and turned off when the value of the first average vector is not "0". Second average vector determination unit 106
Is a circuit for determining whether or not the value of the first average vector H exceeds a predetermined value “α”, and outputs the determination result to the second average vector calculation unit 108 as the selection signal I. Like the signal K, the selection signal I is turned on when the value of the first average vector exceeds "α", and "α"
This signal is turned off in the following cases. The threshold value “α” is set in advance according to the characteristics of the image. For example, if the image is an entirely flat image, the value is set low, and if the image has many gradients, it is set high according to the gradient. .

The second average vector calculation unit 108 determines that the result from the second average vector determination unit 106 is that the first average vector exceeds the predetermined value "α", that is, the selection signal. It is activated when I is turned on, and at that time, the first average vector H supplied from the first average vector calculation unit 102 is sequentially taken in, and this is accumulated over one field, and the average is obtained. Find the value. That is, a value obtained by further averaging the average values of the entire average vectors in the previous field that exceed the predetermined value is supplied to the first initial displacement vector selection unit 110 as the second average vector J.

The first initial displacement vector selection unit 110 is formed by a switching circuit or the like that switches in response to the selection signal K. When the selection signal K is on, that is, the value of the first average vector is "0". , The first average vector H is selected as a candidate for the initial deviation vector, and the selection signal K
Is off, that is, when the value of the first average vector is “0”, the second average vector J is set to the initial deviation vector candidate L.
Is a selection circuit for selecting as.

The second initial displacement vector selection unit 112 reads out the motion vector G of the detected block adjacent to the detected block from the vector memory 41 and receives them from the first initial displacement vector selection unit 110. Average vector L
The selection circuit selects any one of them as the initial displacement vector C. For example, in this embodiment, a difference circuit that performs coordinate conversion of the previous field signal 2 with each initial displacement vector value and takes the difference between the value and the current field signal 1, and this difference value is calculated for each pixel. Includes an absolute value circuit for converting the output to an absolute value, an accumulation circuit for accumulating the output for the number of pixels in the block, and a comparison circuit for comparing each accumulated value. Output as the position vector C.

Next, the motion vector detecting method in this embodiment will be described together with the operation of the motion vector detecting circuit having the above-mentioned configuration. In the operating state, the motion vector D detected by the vector detection unit 50 is sequentially accumulated in the vector memory 41 via the vector evaluation unit 60 for each field. At this time, the first average vector calculation unit 102 sequentially reads the motion vector of the already detected block and the motion vector F of the blocks around it from the vector memory 41, and averages them to obtain the first average vector in block units. Find H. The first average vector H of these sequentially calculated results
Is a first average vector determination unit 104, a second average vector determination unit 106, and a first initial deviation vector selection unit 112.
Is supplied to.

The second average vector determination unit 106, which has received the first average vector H, determines whether or not the first average vector H exceeds a preset threshold value "α", and the determination result is obtained. I is supplied to the second average vector calculation unit 108. In this case, when it is determined that the value of the first average vector H exceeds “α”, the selection signal I from the determination unit 106 is turned on, and the second average vector calculation unit 10
8 is activated, and the vector value H thereof is sequentially fetched into the second average vector calculation unit 108. By repeating this operation, the second average vector calculation unit 108 makes the second determination unit 10
When the first average vector H determined to exceed the value of “α” in 6 is fetched over one field, the calculation unit 108 calculates the average value to obtain one average vector for the entire previous field. The average vector J of 2 is supplied to the first initial displacement vector selection unit 110.

On the other hand, the first receiving the first average vector H
The average vector determination unit 104 determines whether the vector value is “0” and supplies the determination result K to the first initial displacement vector selection unit 110. Then, the determination result K
When the value of the first average vector is not “0”, the first initial displacement vector selection unit 110 that has received the first average vector H as the candidate of the initial displacement vector in the detected block The second initial deviation vector selection unit 11 is selected.
Supply to 2. If the value of the first average vector is “0”, the second average vector J from the second average vector calculation unit 108 is selected and the second initial deviation vector selection unit 112 is selected.
Supply to.

Next, the first initial deviation vector selection unit 11
Mean vector L as a candidate for initial deviation vector from 0
The second initial displacement vector selection unit 110 that has received the data reads out a plurality of motion vectors G of already detected blocks around the detected block from the vector memory 41, and optimizes them as the initial displacement vector of these and the average vector. A different motion vector. In this case, the second initial displacement vector selection unit 110 obtains a signal obtained by coordinate-transforming the previous field signal 2 with each candidate vector, and obtains a difference value between the signal and the current field signal. Next, a value obtained by converting the difference value into an absolute value for each pixel is accumulated in the entire block. When the accumulation is completed for each motion vector, the accumulated values are compared, and the candidate vector having the smallest value, that is, the vector value closest to the motion amount from the previous field to the current field is set as the initial displacement vector C. Selected,
It is supplied to the vector detection unit 50.

Next, the vector detection unit 50 that has received the initial displacement vector C repeats the calculation by the iterative gradient method with the previous field signal displaced by the initial displacement vector C as the starting point, and the true motion is calculated. Detect vector D. The vector value of the motion vector D is evaluated by the vector evaluation unit 60, and if it is included in a predetermined error range, it is output from the output terminal 14 and stored in the vector memory 41.

The above operation is repeated for each block for the current field signal 1, and true motion vectors are sequentially detected for all blocks in the field. Further, when the motion vector for one field is stored in the vector memory 41 by the above operation and the signal of the next field is input, the selection of the initial displacement vector C is executed in the same manner as described above, and each field is selected. The motion vector E is detected every time in the same manner as described above.

As described above, according to this embodiment, the initial deviation vector selection unit 42 is provided with the first average vector determination unit 104.
A second average vector determination unit 106, a second average vector calculation unit 108, and a first initial deviation vector selection unit 110.
Since and are newly provided, the average value over the entire one field of the already detected block is set as the effective value in the portion detected as the motion vector “0” as the second effective value.
The mean vector of can be replaced. As a result, for example, in an image in which most of the area of the image as shown in FIG. 6 is stationary and an object having a small area is moving in the image, from the initial displacement vector of the average vector “0”, It was possible to obtain the result that the vector detection can be performed more accurately when an effective value is given as the initial displacement vector than when the vector detection is performed. In other words, the part that was conventionally selected as the "0" vector or average vector in the vicinity of the boundary between the moving body and the static area, or in a block including a moving body with a small area, is not the above, but the initial deviation closer to that value. The second average vector is selected as the position vector, so that the true motion vector can be effectively detected in a short time.

Further, in this embodiment, since the initial displacement vector can be selected without increasing the number of initial displacement vector candidates, the circuit scale of the second initial displacement vector selection unit 3 does not increase. The effect that the device can be miniaturized is obtained.

In the above embodiment, an interlace type image signal using field signals has been described as an example, but the present invention may be applied to a non-interlace type image signal. In this case, the field signal in this specification may be read as a frame signal. In the above embodiment, when the second initial displacement vector selection unit 112 compares each of the initial displacement vector candidates based on the current field signal and the previous field signal to select the optimum initial displacement vector. However, in the present invention, the second initial displacement vector selection unit compares at least one selected average vector with the motion vector of the detected block and has a large difference from the average vector. Vector comparison means for selecting the motion vector of the already detected block in advance is provided, and the same comparison operation as above is performed for at least two types of the motion vector of the already detected block selected by this comparison means and the selected average vector. You may do it. Further, in the present embodiment, the first average vector exceeding the predetermined value is accumulated, and this is averaged to obtain the second average vector, but in the present invention, it is read from the vector memory. The value of the motion vector of the directly detected block is compared with a threshold value similar to the above predetermined value, those exceeding this value are accumulated in one field, and the averaged value is used as the second average vector. Good.

[0042]

As described in detail above, according to the motion vector detecting method and the motion vector detecting circuit of the moving image signal according to the present invention, when a block whose average vector value is "0" is detected, The average vector of blocks exceeding the value set accordingly is replaced with the second average vector averaged over one field to be the initial deviation vector candidate. Therefore, when this vector is selected as the initial deviation vector, There is an effect that it is possible to effectively detect the motion vector of the detected block having a value close to this vector, for example, the motion vector of the edge of the moving object and the moving vector of the moving object having a small area. In this case, the detection accuracy can be improved without increasing the number of candidates for the initial displacement vector, and at the same time, the size of the device can be reduced.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a motion vector detection circuit to which a motion vector detection method of a moving image signal according to the present invention is applied.

FIG. 2 is a block diagram showing an internal configuration example of a vector evaluation unit according to the embodiment of FIG.

FIG. 3 is a block diagram showing a basic circuit to which a motion vector detection method and a motion vector detection circuit of a moving image signal according to the present invention are applied.

FIG. 4 is a diagram showing a relationship between respective blocks of a current field signal and a previous field signal.

FIG. 5 is a diagram for explaining a motion vector in this embodiment.

FIG. 6 is a diagram showing an example of an image in which a motion vector is detected in the embodiment of FIG.

[Explanation of symbols]

41 Vector memory 42 Initial deviation vector selection section 50 Vector detector 60 Vector evaluation section 102 First average vector calculator 104 First average vector determination unit 106 Second average vector determination unit 108 Second average vector calculator 110 First initial deviation vector selection unit 112 Second initial deviation vector selection unit

─────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP 61-184071 (JP, A) JP 62-206980 (JP, A) JP 4-14980 (JP, A) JP 4- 40185 (JP, A) JP-A-6-153167 (JP, A) JP-A-6-217266 (JP, A) JP-A-7-7721 (JP, A) JP-A-7-135663 (JP, A) JP-A-7-336691 (JP, A) JP-A-8-149421 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) H04N ^7/ 00-7/088 H04N 7/24 -7/68

Claims (9)

(57) [Claims] 1. An initial displacement vector that becomes an initial value when a motion amount is estimated for each block by subdividing each field of a digitized moving image signal into blocks including a plurality of pixels and lines. And a motion vector detection method for a moving image signal in which a motion vector representing a motion amount of each detected block is detected based on the initial displacement vector, the method includes: The first step of accumulating the fields, and the motion vector accumulated in the first step are read out together with the motion vectors of a plurality of blocks around the field, and the average of these values is calculated for each block. The second step of obtaining the average vector and whether or not the value of the first average vector obtained in the second step is "0". And a fourth step of determining whether or not the value of the first average vector obtained in the second step exceeds a predetermined value set in advance for each image. And a first value that exceeds a predetermined value determined in the fourth step.
In the fifth step of obtaining the second average vector by averaging these average vectors over one field and averaging them over the entire field, and the value of the first average vector in the fourth step is "0". When it is determined that the first average vector is selected as a candidate for the initial displacement vector of the detected block, and it is determined that the first average vector is “0”,
A sixth step of selecting the second average vector obtained in the fifth step as a candidate of the initial deviation vector, and a candidate of the initial deviation vector and the detected block selected in the sixth step Receiving a motion vector of an already detected block adjacent to, and selecting a motion vector having a value closest to the value of the motion vector of the detected block as an initial displacement vector, and the seventh step. The eighth step of receiving the initial displacement vector selected in step 1, the image signal of the current field, and the image signal of the previous field, and detecting the true motion vector of the detected block based on these signals. Motion vector detection method for moving image signals. 2. The motion vector detecting method according to claim 1, wherein in the seventh step, an image signal of a block of a previous field corresponding to a detected block is obtained with each value of candidates of the initial displacement vector. A step of coordinate conversion, a step of obtaining a difference value between the coordinate-converted signal and the image signal of the detected block for each pixel, a step of accumulating the absolute value of the obtained difference value by the number of pixels of the block, A method of detecting a motion vector of a moving image signal, comprising the step of comparing respective accumulated values and selecting a motion vector having a minimum accumulated value as an initial displacement vector. 3. The motion vector detection method according to claim 2, wherein the seventh step is to select one of the average vectors selected in the sixth step and a detected block adjacent to the detected block. Comparing with the motion vector, selecting at least one motion vector of the detected block having a large difference from the average vector, and selecting the vector and the average vector as candidates for the initial displacement vector. Characteristic motion vector detection method. 4. The motion vector detecting method according to claim 1, wherein the eighth step receives the image signals of the previous field and the current field, and selects the image signal of the previous field in the seventh step. By using the signal displaced by the initial displacement vector as a starting point, the amount of motion displacement with respect to the image signal of the current field is obtained by repeatedly performing a plurality of gradient calculations based on the gradient amount and the difference amount of each signal, A method for detecting a motion vector of a moving image signal, which comprises detecting a true motion vector of a detected block. 5. An initial displacement vector which is an initial value for estimating a motion amount for each block by subdividing each field of the digitized image signal into blocks including a plurality of pixels and lines. In a motion vector detection circuit that sets and detects a motion vector representing a motion amount of each detected block based on the initial displacement vector, the circuit stores a motion vector of an already detected block for at least one field. Means, and a first average vector operation means for reading out the motion vector accumulated in the storage means together with the motion vectors of a plurality of blocks around the storage means, and obtaining a first average vector by averaging the motion vectors for each block. Whether or not the value of the first average vector calculated by the first average vector calculation means is "0" And a first determination unit for determining whether or not the value of the first average vector obtained by the first average vector calculation unit exceeds a predetermined value preset according to the image. A second judging means for judging and a first average vector exceeding a predetermined value judged by the second judging means are accumulated over one field, and these are averaged over the entire field.
Of the initial deviation vector of the detected block when the value of the first average vector is not “0” by the second average vector calculation means for obtaining the average vector of When the first average vector is selected as a candidate and it is determined that the first average vector is “0”, the second average vector calculation unit obtains the second average vector.
Selection means for selecting the average vector of the above as a candidate for the initial displacement vector, and the candidate for the initial displacement vector selected by the first selection means and the motion vector of the detected block adjacent to the detected block In response to this, second selection means for selecting a motion vector having a value closest to the value of the motion vector of the detected block as the initial displacement vector, and the initial displacement selected by the second selection means. A motion vector detection circuit comprising: a vector, a current field image signal, and a previous field image signal, and vector detection means for detecting the true motion vector of the detected block based on these signals. 6. The motion vector detection circuit according to claim 5, wherein the second selection means is an image signal of a block in the previous field corresponding to the detected block with each value of the candidates for the initial displacement vector. Transforming means for coordinate transforming, a difference generating means for obtaining a difference value between the coordinate-transformed signal and the image signal of the detected block for each pixel, and an absolute value of the difference value is accumulated for the number of pixels of the block. Accumulating means, comparing means for comparing outputs of the accumulating means, and selecting means for receiving a result of the comparing means and selecting a motion vector having a minimum accumulated value as an initial displacement vector Motion vector detection circuit. 7. The motion vector detection circuit according to claim 6, wherein the second selection unit moves the average vector selected by the first selection unit and a detected block adjacent to the detected block. And a vector comparison means for comparing at least one motion vector of the detected block having a large difference from the average vector with the motion vector of the detected block selected by the vector comparison means. A motion vector detecting circuit, characterized in that one of the average vectors selected by the first selecting means is selected as an initial displacement vector. 8. The motion vector detecting circuit according to claim 5, wherein the vector detecting means receives the image signals of the previous field and the current field, and selects the image signal of the previous field by the second selecting means. And a plurality of gradient calculating means for calculating the amount of motion deviation with respect to the image signal of the current field based on the gradient amount and the difference amount of each signal with the signal displaced by the initial displacement vector as a starting point. Characteristic motion vector detection circuit. 9. The motion vector detection circuit according to claim 5, wherein the circuit evaluates whether or not the value of the motion vector detected by the vector detection means is detected within a predetermined error range. Including only the evaluation means, outputting only the motion vector which is good by the vector evaluation means,
And a motion vector detection circuit which is stored in the storage means.