JP2007097028A - Motion vector detecting method and motion vector detecting circuit - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a high-precision motion vector detecting method and a motion vector detecting circuit by which a correct motion vector is easily detected in a still region. <P>SOLUTION: In the motion vector detecting circuit 10, a motion detecting circuit 22 which receives a reference field (preceding field) signal 16 and a detected field (current field) signal 20, calculates a total sum ΣFD of field difference between a detected block within a detected field at the same position and a reference block within a reference field before detecting a motion vector between these fields. A motion determination circuit 20 determines whether the detected block is moving/still on the basis of the total sum ΣFD of the field difference. If a determination result 32 is "still", an initial bias vector selection circuit 42 selects an initial bias vector, improving a priority order of a zero vector, from among a plurality of vectors. A subsequent gradient method arithmetic circuit 52, 54 detects a motion vector V on the basis of the selected initial bias vector. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a method and circuit for detecting the size and direction of a moving image, that is, a motion vector, in a TV (Tele-Vision) signal.

  The motion vector is used when improving the inter-frame coding efficiency in high-efficiency coding of a TV signal, or when reducing motion discontinuity due to the conversion of the number of fields in the TV system conversion.

  A general motion vector detection method is a method of detecting a motion vector for each block after subdividing a TV signal into blocks of m pixels × n lines (m and n are integers). The pattern matching method disclosed in Patent Document 3 and the iterative gradient method disclosed in Patent Document 4 are well known.

  Furthermore, some iterative gradient methods use an initial deviation vector (Patent Document 5) in order to improve motion vector detection accuracy. In this method, a motion vector is selected from among a plurality of motion vectors that have already been detected as an initial deviation vector, and a predetermined deviation vector is detected using this initial deviation vector as a starting point, and a true motion vector is obtained. This is a detection method.

  Patent Document 1 is also a technique that uses an iterative gradient method by selecting an initial deviation vector. Patent Document 1 is characterized by processing after detecting a motion vector by the iterative gradient method. In other words, when a static region that is stationary from the previous field is detected in the motion region where the motion vector is detected, the surrounding motion vector is replaced with a zero vector. .

Japanese Patent No. 3121519 Japanese Unexamined Patent Publication No. 55-162683 Japanese Patent Laid-Open No. 55-162684 JP 60-158786 A Japanese Patent Laid-Open No. 62-206980

  For example, in Patent Document 1, the vectors prepared as candidates for the initial deviation vector are a motion vector detected in the past in the vicinity of the detected block whose motion vector is to be detected, and a zero vector. Usually, the priority order of neighboring motion vectors as initial deviation vector candidates is high so that the motion vector to be detected in the detected block follows the neighboring motion. Therefore, when a moving part and a stationary part are mixed in the video signal, the stationary part where the zero vector should be detected as a motion vector is influenced by the moving part in the vicinity. There is a risk that a certain amount of motion vectors will be erroneously detected. As a result, there is a drawback that the contour and shape of the stationary part are disturbed.

  In view of such a problem, an object of the present invention is to provide a highly accurate motion vector detection method and circuit that can easily detect a correct motion vector even in a stationary region.

  In order to solve the above-described problems, the present invention subdivides a digitized video signal into blocks of a predetermined size for each field, and detects one or more fields from the detected field for the detected block in the detected field. In a motion vector detection method for detecting a motion vector based on a distant reference field, a sum of differences between fields between a detected block and a reference block at the same position as the detected block in the reference field , A step of determining the motion / stillness of the detected block based on the sum of the differences between fields, and an initial deviation that is an initial value for detecting a motion vector when it is determined that the motion block is still in this determination step Among the multiple vectors that are candidate vectors, the initial vector is determined by increasing the priority of the zero vector. A step of-option, and a step of detecting a motion vector based on the initial vector selected in this step, thereby characterized in that to improve the accuracy of motion vector detection.

  According to the present invention, when a moving part and a stationary part are mixed in the video, the stationary part is affected by the surrounding moving part and erroneously detects a motion vector. The disadvantage of disturbing the contour and shape of the stationary part can be eliminated.

  Next, embodiments of a motion vector detection method and a motion vector detection circuit according to the present invention will be described in detail with reference to the accompanying drawings. In the present text, illustration and description of elements not directly related to the present invention are omitted. In addition, the signal is indicated using the sign of the signal line on which the signal appears.

  FIG. 1 is a block diagram showing an embodiment of a motion vector detection circuit according to the present invention. The motion vector detection circuit 10 is a circuit that receives a digitized video signal as an input and detects a motion vector. The motion vector is detected based on the preceding reference field (previous field) among the continuous fields constituting the video signal, and the motion to the subsequent detected field (current field) one or more fields away from the reference field. Do. More specifically, motion vector detection is performed for each block by dividing the detected block into blocks of a predetermined size.

  As shown in FIG. 1, the motion vector detection circuit 10 includes a field delay circuit 12, and this circuit 12 is connected to an input terminal 14 of the motion vector detection circuit 10. The field delay circuit 12 is a circuit that delays the video signal 14 input thereto by one field and outputs a reference field signal 16 that has been subjected to processing such as amplification as necessary. An amplifier 18 is also connected to the input terminal 14 of the motion vector detection circuit 10. In this way, the video signal 14 is also output as the detected field signal 20 that has been subjected to only amplification processing that is not delayed in time by the amplifier 18. The reference field signal 16 becomes a reference for motion vector detection as a preceding field. On the other hand, the detected field signal 20 is a target of motion vector detection.

  As shown in FIG. 1, the motion vector detection circuit 10 includes a motion detection circuit 22, which receives a reference field signal 16 and a detected field signal 20 as input signals. The motion detection circuit 22 is a circuit that subdivides the reference field and the detected field into blocks of a predetermined size. The block size may be, for example, 8 pixels × 8 lines, or any other block size. The motion detection circuit 22 also calculates the sum ΣFD of the field difference FD between the detected block in the detected field and the reference block at the same position as the detected block in the reference field. And a circuit for outputting.

  Here, the difference between fields will be described. FIG. 2 is a diagram showing a block correspondence between a reference field and a detected field input to the motion detection circuit 22 in FIG. As shown in FIG. 2, the detected field 20 includes a detected block 26 whose motion vector should be detected with reference to the reference field 16. In the reference field 16, there is a reference block 28 at the same position as the detected block 26. The inter-field difference FD is a signal difference for each pixel at a corresponding position between the detected block 26 and the reference block 28 as shown in FIG. The motion detection circuit 22 adds the inter-field difference FD over the entire block, and calculates the sum ΣFD of inter-field differences.

  As shown in FIG. 1, the motion vector detection circuit 10 includes a motion determination circuit 30 that is connected to the output ΣFD of the motion detection circuit 22 and receives the sum ΣFD of inter-field differences as an input signal. The motion determination circuit 30 is a circuit that determines the motion / stillness of the detected block based on the sum ΣFD of the inter-field differences and outputs the determination result as a signal 32.

  The motion determination circuit 30 determines a motion when the inter-field difference sum ΣFD is larger than a predetermined threshold. That is, the reference block 28 and the detected block 26 at the same position are different images. Therefore, the detected block 26 is different from the reference block 28 and is moved from another block in the reference field 16. judge. On the other hand, when the sum ΣFD of the inter-field differences is smaller than the predetermined threshold, in other words, the motion determination circuit 30 determines to be still when it is zero or approximate to zero with a certain error. That is, there is a high possibility that the reference block 28 and the detected block 26 at the same position are the same image. Therefore, the detected block 26 determines that the reference block 28 is kept still.

  As shown in FIG. 1, the motion vector detection circuit 10 includes block delay circuits 34 and 36. The circuit 34 receives the reference field signal 16 as an input signal, and the circuit 36 receives the detected field signal 20 as an input signal. To do. The block delay circuits 34 and 36 are circuits that subdivide the input field signals 16 and 20 into blocks of a predetermined size, delay them, and output them as block signals 38 and 40, respectively. The size of the block is the same as that in which the motion detection circuit 22 divides the field signal into blocks.

  The detected block signal 38 and the reference block signal 40 are input to the circuit 42 from the block delays 34 and 36, and the motion / stillness determination result 32 is obtained from the motion determination circuit 30 to the circuit 42 for the detected block / reference block. You may synchronize with the timing inputted. In any case, it is only necessary that the block signals 38 and 40 of the entire fields are input to the circuit 42 when the motion / stillness determination result 32 is input to the circuit 42 over the entire reference field / detected field.

As shown in FIG. 1, the motion vector detection circuit 10 includes a vector memory 44. The vector memory 44 is a storage device that stores a plurality of vectors. The plurality of vectors stored in the vector memory 44 are candidates for an initial displacement vector V ₀ described later. FIG. 3 is a diagram showing an example of initial deviation vector candidates stored in the vector memory 44 of FIG. Symbols A to C, F to N, and P to X all indicate motion vectors in the respective blocks in the continuous fields 48, 16, and 20 that have already been detected in the past by gradient method arithmetic circuits 52 and 54 described later. .

  As shown in FIG. 3, in the vector memory 44, motion vector A to already detected for the detected block 26 indicated by the hatched area, that is, the block in the vicinity of the block from which the motion vector is to be detected will be stored as initial deviation vector candidates. C may be included. Among these candidates, motion vectors A and B are motion vectors detected in blocks belonging to the vicinity of 8 of the detected block 26. Further, even if a motion vector is detected in the past in a block that does not belong to the vicinity of the detected block 26, such as the motion vector C, the motion vector is detected in a block in the vicinity of a predetermined range. The initial deviation vector candidate may be included in the vector memory.

  FIG. 3 shows a reference field 16 that is one field ahead of the detected field 20, and a field 48 that is one field ahead. Needless to say, there is a preceding field before the field shown in FIG. Such motion vectors F to N and P to X that have already been detected in a field one field or more away from the detected field 20 may also be included in the vector memory 44 as candidates for the initial displacement vector.

As shown in FIG. 1, the motion vector detection circuit 10 includes an initial deviation vector selection circuit 42, and the circuit 42 receives a motion / still signal 32, which is an output of the motion determination circuit 30, as an input signal. The initial deviation vector selection circuit 42 is also connected to the output terminal 46 of the vector memory 44. The initial deviation vector selection circuit 42 assigns priorities to a plurality of vectors received from the vector memory 44, further changes the above priorities according to the motion / still signal 32, and performs initial deviation in order from the higher order. This is a circuit that examines whether or not the position vector V ₀ is suitable, and if it finds a suitable one, selects it as the initial displacement vector V ₀ . Here, the “initial deviation vector” is an initial value for detecting the motion vector V to be finally detected.

Candidates for the initial displacement vector V _{0 that} the initial displacement vector selection circuit 42 reads from the vector memory 44 belong to the vicinity of the detected block 26 and other neighborhoods like the motion vectors A to C shown in FIG. A motion vector detected in the past in the block may be used. If the number of candidates is still insufficient, motion vectors F to N detected in the past in the reference field 16 are added in order. If there is still a shortage, motion vectors detected in the past in the field 48 are added. P to X may be added. When selecting a candidate of the initial displacement vector V ₀ from a preceding field that is one field or more away from the detected field 20 such as the reference field 16 or the previous field 48, first, as in the reference block 28 of the reference field 16 The motion vector detected in the past in the block at the same position as the detected block 26, that is, the vector J, may be preferentially added to the candidate, and then the neighboring motion vector may be added to the candidate. Note that the motion vectors already detected in the past shown in FIG. 3 may be freely added and averaged to be a candidate for the initial displacement vector V ₀ .

The number of candidates for the initial displacement vector V _{0 that} the initial displacement vector selection circuit 42 reads from the vector memory 44 may be eight, for example, but may be any number greater or smaller than this. However, a zero vector is always included in the candidates. Therefore, when the number of candidates is 8, the initial deviation vector selection circuit 42 reads seven candidate vectors from the vector memory 44, and sets the remaining one candidate as a zero vector.

The initial displacement vector selection circuit 42 may set the priority order of a plurality of vectors that are candidates for the initial displacement vector V ₀ excluding the zero vector as the descending order of the absolute value of the vector. This is to make it easier to detect a motion vector that follows a motion vector as large as possible detected in the vicinity.

When the motion / still signal 32 is “still”, the initial displacement vector selection circuit 42 increases the priority of the zero vector among the plurality of vectors that are candidates for the initial displacement vector V ₀ , and performs the initial displacement. select a vector _{V 0.} The priority of the zero vector may be any number as long as it is higher than the previous one and may be the highest.

On the other hand, when the motion / still signal 32 is “motion”, the initial displacement vector selection circuit 42 lowers the priority of the zero vector among a plurality of vectors that are candidates for the initial displacement vector V ₀ , The deviation vector V ₀ is selected. The priority of the zero vector may be any number as long as it is lower than the priorities, and may be the lowest. If the zero vector has the lowest priority from the beginning, it may be maintained as it is.

  In this embodiment, the motion / still signal 32 is one of “static” and “motion”. However, the motion determination circuit 30 described above may output signals other than these two as determination results. . For example, three or more types of signals may be output as determination results in accordance with the magnitude of the inter-field difference sum ΣFD. In that case, the initial deviation vector selection circuit 42 may set the priority order of the zero vectors to any one of three or more according to the determination result.

Initial vector selection circuit 42, a plurality of vectors which are candidates for initial vector V ₀ and the candidate vector in order of priority, to consider whether or not that may or initial vector V ₀ . This examination may be performed as follows. That is, first, the candidate vector is assumed to be the initial displacement vector V ₀ . Then, an inter-field difference between the detected block 26 shown in FIG. 2 and the corresponding block 50 located in the reference field 16 and before the detected block is displaced by the initial displacement vector V ₀ . Calculate DFD (Displaced Field Difference) sum ΣDFD.

FIG. 4 is a diagram showing a difference between the sum ΣFD of inter-field differences FD calculated by the motion detection circuit 22 shown in FIG. 1 and the sum ΣDFD of inter-field differences DFD calculated by the initial deviation vector selection circuit 42. As shown in FIG. 4, the sum ΣFD of the inter-field difference FD is calculated between the detected block 26 and the reference block 28 in the reference field 16 at the same position. On the other hand, the sum ΣDFD of the inter-field difference DFD is obtained before the detected block 26 is displaced by the detected block 26 and the initial displacement vector V ₀ assumed in the reference field 16 as shown in FIG. Is calculated between the corresponding block 50 and the corresponding block 50. The relationship between the corresponding block 50 and the detected block 26 is also shown in FIG. The coordinates of the corresponding block 50 are (m ₁ , n ₁ ), the initial displacement vector V ₀ = (α ₀ , β ₀ ), and the coordinates (m ₁ ) where the corresponding block 50 is displaced by the initial displacement vector. + Α ₀ , n ₁ + β ₀ ) is the detected block 26.

If the sum ΣDFD of the inter-field difference DFD is smaller than a predetermined threshold, in other words, if it is zero or approximates to zero with a certain error, the candidate vector is formally selected as the initial displacement vector V ₀ . Alternatively, a candidate vector that minimizes the sum ΣDFD is formally selected as the initial displacement vector V ₀ . When the sum ΣDFD of the inter-field differences has the same minimum value due to a plurality of candidates, a candidate having a predetermined high priority is selected as the initial deviation vector. The difference between the video signal of the corresponding block 50 and the video signal of the detected block 26 is small, and both are similar, and it is assumed that the detected block 26 was at the position of the corresponding block 50 in the reference field 16 Because it can.

As shown in FIG. 1, the motion vector detection circuit 10 includes a gradient method calculation circuit 52, these circuits 52 and 54, relative to the initial vector V _0, a circuit for applying the iterative gradient method. Since the configurations and processes of the circuits 52 and 54 are common, the circuit 52 will be described below as a representative. The gradient method computing circuit 52 applies the gradient method to the initial displacement vector V ₀ that is the output of the initial displacement vector selection circuit 42. The first gradient method computing circuit 52 performs the calculations shown in the following equations (1) to (4) on the block before the detected block 26 is displaced, that is, the corresponding block 50 and the detected block 26. no rows, an arithmetic circuit for obtaining a first shift vector V _1.
V _X = Σ {(SGNΔX) · DFD} / Σ | ΔX | (1)
V _Y = Σ {(SGNΔY) · 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 the value V _X is the X-direction component of the _first displacement vector V ₁ , the value V _Y is the Y-direction component of the _first displacement vector V ₁ , the values _{An, m} are n pixels, and the m-line coordinate signal, The value ΔX represents the gradient in the X direction of the image, the value ΔY represents the gradient in the Y direction of the image, and the value DFD represents the inter-field difference value. SGN indicates the signs of ΔX and ΔY including “0”.

The first deviation vector V ₁ is added to the initial deviation vector V ₀ by the adder 56 to become a vector V ₀ + V ₁ and input to the next second gradient method computing circuit 54. Similarly to the first gradient method calculation circuit 52, the circuit 54 performs the second gradient method calculation on the vector V ₀ + V ₁ and outputs the second displacement vector V ₂ . The output second displacement vector V ₂ is added to the vector V ₀ + V ₁ by the adder 58, and a motion vector V = V ₀ + V ₁ + V ₂ to be finally detected is detected. FIG. 5 is a diagram for explaining the relationship between these motion vectors. The detected motion vector V is input to the vector memory 44 as a motion vector already detected in the past, and is used as a candidate for the initial displacement vector V ₀ when detecting a motion vector for another block.

The advantages of the present invention will now be described. The ultimate object of the present invention is to detect the motion vector V for the detected block 26. However, if the detected block 26 is originally an area that has not moved between the reference field 16 and the detected field 20, the motion vector to be finally detected is a zero vector. Therefore, before detecting a motion vector by the later-described iterative gradient method using the initial displacement vector V ₀ selected by the initial displacement vector selection circuit 42, the motion detection circuit 22 / motion determination shown in FIG. A circuit 30 is provided to determine whether or not the detected block 26 is originally stationary. If the determination result is “still”, the priority of the zero vector is increased among the candidates for the initial displacement vector V ₀ , and it is preferentially examined whether this can be the initial displacement vector.

In this way, since it is determined that the detected block is originally stationary, there is a high possibility that the zero vector is selected as the initial deviation vector V ₀ . The motion vector finally detected by the iterative gradient method is also likely to be a zero vector or a vector close thereto, and it is an advantage of the present invention that the accuracy of motion vector detection is improved.

FIG. 6 is a block diagram showing an example of a motion vector detection circuit using a conventional initial deviation vector. A feature of the configuration of the embodiment of the present invention shown in FIG. 1 is that a motion detection circuit 22 and a motion determination circuit 30 are added to the conventional motion vector detection circuit 100 of FIG. Because of these structural features, the use of the embodiment of the present invention is more likely to be selected as an appropriate initial displacement vector V ₀ than the circuit 100 of FIG.

  The operation of the embodiment of the motion vector detection circuit according to the present invention having the configuration described so far will be described below. When the video signal 14 is input from the input terminal 14 of the motion vector detection circuit 10 shown in FIG. 1 to the field delay circuit 12, the circuit 12 delays the video signal 14 by one field and performs processing such as amplification as necessary. The performed reference field signal 16 is input to the motion detection circuit 22 and the block delay circuit 34. On the other hand, the video signal 14 is converted into a detected field signal 20 that has undergone only amplification processing that is not delayed in time by the amplifier 18, and is similarly input to the motion detection circuit 22 and the block delay circuit 34. Thus, the reference field signal 16 is a signal that is one field past the detected field signal 20.

  Using the received signals 16 and 20, the motion detection circuit 22 subdivides the reference field and the detected field into blocks of a predetermined size. The motion detection circuit 22 also calculates a sum ΣFD of inter-field differences FD (Field Difference) between the detected block in the detected field and all the reference blocks at the same position in the reference field as the detected block. calculate. The calculation of the total sum ΣFD is performed for each block over all the blocks of the reference field and the detected field, and is input to the motion determination circuit.

  The motion determination circuit 30 determines the motion / stillness of the detected block over the entire detected field based on the total difference ΣFD between the fields, and inputs the determination result as a signal 32 to the initial displacement vector selection circuit 42.

  The block delay circuits 34 and 36 use the received reference field signal 16 and the detected field signal 20, respectively, to subdivide the fields into blocks of a predetermined size, delay them, and block signals 38 for each block, The initial deviation vector selection circuit 42 is input as 40. Thus, when the motion / stillness determination result 32 is input to the circuit 42 over the entire reference field / detected field, the input of the block signals 38 and 40 of the entire field to the circuit 42 is completed.

  The initial deviation vector selection circuit 42 reads a predetermined plurality of, for example, seven vectors from the vector memory 44 when selecting an initial deviation vector as an initial value for detecting a motion vector of a detected block. A total of 8 candidate vectors are prepared by adding a zero vector. Priorities are assigned to these candidate vectors. The priority may be set in descending order of the absolute value of the vector, for example. Further, if the determination result of the motion / still signal 32 of the detected block is “still”, the priority of the zero vector among the candidates for the initial deviation vector is increased. At this time, the zero vector may be the highest. On the other hand, if the determination result is “motion”, the priority of the zero vector is lowered.

The initial displacement vector selection circuit 42 assumes that each candidate vector is an initial displacement vector V ₀ according to the priority order determined as described above. Then, the sum ΣDFD of the inter-field difference DFD (Displaced Field Difference) is calculated between the detected block 26 and the corresponding block 50 shown in FIG. If the sum ΣDFD of the inter-field difference DFD is smaller than a predetermined threshold, in other words, if it is zero or approximates to zero with a certain error, the candidate vector is formally selected as the initial displacement vector V ₀ . Alternatively, a candidate vector that minimizes the sum ΣDFD is formally selected as the initial displacement vector V ₀ . The officially selected initial deviation vector V ₀ is input to the gradient method arithmetic circuit 52.

The gradient method arithmetic circuits 52 and 54 apply the iterative gradient method to the initial displacement vector V ₀ , the circuit 52 calculates the _first displacement vector V ₁ , and the circuit 54 performs the second displacement vector V _2. Is calculated. These deviation vectors are added, and finally a motion vector V = V ₀ + V ₁ + V ₂ is detected. Note that the motion vector V detected in this way is input to the vector memory 44 as a motion vector already detected in the past, and the initial displacement vector V ₀ when detecting a motion vector for another block is obtained. Used as a candidate.

  The motion vector detection circuit according to the present invention detects a motion vector from an input video signal. Therefore, it can be incorporated in a TV system conversion device. For example, when a TV signal is converted from a PAL system to an NTSC system by a TV system conversion device, it is necessary to convert the number of fields from 50 fields to 60 fields. Therefore, the number of fields is converted by forming one interpolation field for every five fields. In such a case, the motion vector detection circuit according to the present invention can form an interpolation field by detecting a motion vector. The present invention is also applicable to other devices that require motion vector detection, such as a TV signal compression / decompression device (MPEG2-CODEC).

It is a block diagram which shows the Example of the motion vector detection circuit by this invention. It is a figure which shows the correspondence of the block of the reference | standard field input into the motion detection circuit shown in FIG. 1, and a to-be-detected field. It is a figure which shows the example of the candidate of an initial deviation vector memorize | stored in the vector memory shown in FIG. It is a figure which shows the difference between the sum total of the difference between fields which the motion detection circuit shown in FIG. 1 calculates, and the sum total of the difference between fields which an initial deviation vector selection circuit calculates. It is a figure which shows the relationship between the motion vector which should be finally detected, and the displacement vector of the 1st time and the 2nd time. It is a block diagram which shows the example of the motion vector detection circuit using the conventional initial deviation vector.

Explanation of symbols

10 Motion vector detection circuit
22 Motion detection circuit
30 Motion judgment circuit
42 Initial deviation vector selection circuit
44 vector memory

Claims (20)

The digitized video signal is subdivided into blocks of a predetermined size for each field, and a motion vector based on a reference field one or more fields away from the detected field is detected for the detected block in the detected field. In the motion vector detection method, the method includes:
Calculating the sum of inter-field differences between the detected block and a reference block at the same position as the detected block in the reference field;
Determining movement / stillness of the detected block based on a sum of differences between the fields;
If it is determined that the motion vector is stationary in the determining step, among the plurality of vectors that are candidates for the initial displacement vector, which is an initial value for detecting the motion vector, the priority of the zero vector is increased and the initial displacement vector is selected. And a process of
Detecting a motion vector based on the initial deviation vector selected in the step,
A motion vector detection method characterized in that the accuracy of motion vector detection is thereby improved.   The method according to claim 1, wherein in the step of determining the motion / stillness of the detected block, the motion is determined when the sum of the inter-field differences is larger than a predetermined threshold, and is stationary when the sum is smaller than the predetermined threshold. A motion vector detection method, characterized by:   The method according to claim 1 or 2, wherein, in the step of determining, when it is determined to be stationary, the step of selecting the initial deviation vector sets the priority of the zero vector to the highest order. Detection method.   4. The method according to claim 1, wherein the method further includes lowering the priority of a zero vector among the plurality of vectors when the motion is determined in the determining step. A motion vector detection method comprising a step of selecting.   5. The motion vector detection method according to claim 1, wherein the plurality of vectors include motion vectors that have already been detected for blocks in the vicinity of the detected block.   6. The motion vector detection method according to claim 5, wherein the neighboring blocks include blocks belonging to eight neighborhoods of the detected block.   7. The method according to claim 1, wherein the plurality of vectors include motion vectors that have already been detected for a block of a field that is one field or more away from the detected field. Detection method.   8. The motion vector detection method according to claim 1, wherein the priority of the plurality of vectors excluding the zero vector is set in descending order of the absolute value of the vector.   9. The method according to claim 1, wherein in the step of selecting the initial deviation vector, the plurality of vectors are set as candidate vectors in order from the highest priority, the detected block, and the reference The sum of differences between fields is calculated with a corresponding block in the field and before the detected block is displaced by the candidate vector, and a candidate vector that makes the sum smaller than a predetermined threshold is calculated A motion vector detection method comprising selecting an initial deviation vector or selecting a candidate vector that minimizes the sum as the initial deviation vector.   10. The motion vector detection method according to claim 1, wherein in the step of detecting the motion vector, a motion vector is detected by applying an iterative gradient method to the initial deviation vector. . The digitized video signal is subdivided into blocks of a predetermined size for each field, and a motion vector based on a reference field one or more fields away from the detected field is detected for the detected block in the detected field. In the motion vector detection circuit, the circuit includes:
Means for calculating a sum of differences between fields between the detected block and a reference block at the same position as the detected block in the reference field;
Means for determining movement / stillness of the detected block based on a sum of differences between the fields;
Storage means for storing a plurality of vectors that are candidates for an initial deviation vector that is an initial value for detecting the motion vector;
A means for increasing the priority of a zero vector among the plurality of vectors and selecting the initial displacement vector when the determining means determines that it is stationary;
Means for detecting a motion vector based on the initial deviation vector selected by the means,
Thus, a motion vector detection circuit characterized by improving the accuracy of motion vector detection.   12. The circuit according to claim 11, wherein the means for determining movement / stillness of the detected block is determined to be motion when the sum of the inter-field differences is larger than a predetermined threshold, and is stationary when smaller than a predetermined threshold. A motion vector detection circuit characterized by:   13. The motion vector according to claim 11 or 12, wherein, when the determining means determines that it is stationary, the means for selecting the initial displacement vector sets the priority of the zero vector to the highest priority. Detection circuit.   The circuit according to any one of claims 11 to 13, wherein the means for selecting the initial deviation vector lowers the priority of the zero vector among the plurality of vectors when the determination unit determines that the movement is performed. A motion vector detecting circuit for selecting the initial deviation vector.   15. The motion according to claim 11, wherein the plurality of vectors include a motion vector that has already been detected by the means for detecting the motion vector for a block in the vicinity of the detected block. Vector detection circuit.   16. The motion vector detection circuit according to claim 15, wherein the means for selecting the initial displacement vector includes blocks belonging to eight neighborhoods of the detected block in the neighboring blocks.   17. The circuit according to claim 11, wherein the plurality of vectors include a motion vector that is detected by the means for detecting the motion vector for a block in a field that is one field or more away from the detected field. A motion vector detection circuit characterized by the above.   The circuit according to any one of claims 11 to 17, wherein the means for selecting the initial deviation vector sets the priority of the plurality of vectors excluding the zero vector as descending order of the absolute value of the vector. A motion vector detection circuit.   The circuit according to any one of claims 11 to 18, wherein the means for selecting the initial deviation vector sets the plurality of vectors as candidate vectors in descending order of priority, the detected block, and the reference The sum of differences between fields is calculated with a corresponding block in the field and before the detected block is displaced by the candidate vector, and a candidate vector that makes the sum smaller than a predetermined threshold is calculated A motion vector detection circuit which selects as an initial deviation vector or selects a candidate vector which minimizes the sum as the initial deviation vector.   20. The motion vector detection circuit according to claim 11, wherein the motion vector detecting unit detects a motion vector by applying an iterative gradient method to the initial deviation vector. .

Images