JP4883029B2 - Motion vector detection circuit, motion vector detection device, and integrated circuit - Google Patents

Description

  The present invention relates to a technique for detecting a motion vector from an image signal, and more particularly to a motion vector detection circuit, a motion vector detection device, and an integrated circuit incorporating the function, which improve motion vector detection accuracy.

  As a conventional motion vector detection device for an image signal, there is a configuration based on a block matching method. That is, the target image is divided into a plurality of blocks, and the degree of correlation between the target block whose motion is to be detected and a plurality of candidate regions (candidate blocks) within a predetermined search range in the frame before or after the target image is calculated. Each of the candidate blocks is evaluated to determine a candidate block having the highest correlation, and the displacement between the candidate block and the target block is used as a motion vector.

  Further, as one of the plurality of candidate blocks described above, by using the average result of the detected motion vectors in the block in the line direction including the target block or the block in the column direction including the target block, Some configurations improve detection performance.

In addition, when the target block is a boundary block of an image, by assigning a representative value of the detected motion vector to the block in the line direction including the target block or the block in the column direction including the target block. Some configurations improve detection performance.
Japanese Patent Laying-Open No. 2005-287047 JP 2005-287048 A

  In motion vector detection of an image signal, there is a problem that a correct motion vector that should be detected cannot be detected effectively. For example, in an image in which a plurality of objects with different motions are mixed such that the object and the background move in different directions, the above-described problem is likely to occur in the boundary region of the moving object due to the plurality of motions.

  An object of the present invention is to solve the above-described conventional problems and to effectively detect a correct motion vector.

  To achieve the above object, the present invention estimates a motion vector of a reference block, which is a partial region of a display image constituting an image signal, in a motion vector detection circuit that detects a motion vector from an input image signal. An estimated motion vector, a motion vector estimator that calculates a correlation value between the estimated block indicated by the estimated motion vector and the reference block, and a first calculated when the motion vector estimator is applied to the first reference block A motion vector conversion unit that calculates a second reference block according to the estimated motion vector, and a second correlation value that is a correlation value calculated when the motion vector estimation unit is applied to the second reference block Corresponding to the first correlation value and the second correlation value calculated when the correlation value calculation unit and the motion vector estimation unit are applied to the first reference block , Characterized in that it comprises a motion vector determination unit that generates and outputs a motion vector of a second reference block.

  Note that the motion vector conversion unit performs a second reference block on a block at a position indicated by a vector having substantially the same direction or a substantially opposite direction as the first estimated motion vector, and further the same absolute value as the first estimated motion vector. The block at the position indicated by the vector having the magnitude of the value may be used as the second reference block.

  In addition, the motion vector conversion unit may calculate a plurality of second reference blocks for one first estimated motion vector.

  Further, the motion vector conversion unit outputs a block at a position indicated by a vector obtained by adding a vector having a certain size to the first estimated motion vector or a vector obtained by multiplying the first estimated motion vector by a certain value. The reference block may be used.

  In addition, the motion vector determination unit may determine a motion vector by comparing the first correlation value with a certain threshold value, or conversely, determine a motion vector by comparing the second correlation value with a certain threshold value. Also good.

  The present invention relates to a module that is a partial function of a device such as an electric circuit or an electronic circuit, a processing device such as a motion vector detection device for the function, or a hardware such as an integrated circuit incorporating the function. It can be realized not only as a hardware configuration but also as software operating on an arithmetic device such as a CPU, MPU, DSP, or the like.

  An object of the present invention is to detect a motion vector with higher accuracy.

Hereinafter, examples relating to embodiments of the present invention will be described with reference to the drawings. In addition, all the numbers described in the following are illustrated in order to specifically explain the present invention, and the present invention is not limited to the illustrated numbers.
(First embodiment)
FIG. 1 is a functional block diagram illustrating a configuration of the motion vector detection device according to the first embodiment. The motion vector detection apparatus according to the first embodiment inputs an RGB signal as an image signal such as a moving image or a still image, and outputs a motion vector. The motion vector detection apparatus 100 includes a YUV conversion unit 101, a motion vector estimation unit 102, a motion vector conversion unit 103, a correlation value calculation unit 104, and a motion vector determination unit 105.

  The YUV conversion unit 101 converts an input RGB color space image signal into a YUV color space image signal. The YUV conversion unit 101 is inserted in order to maintain consistency between the input image signal and the signal input to the subsequent processing, and the inputs of the motion vector detection device 100 and the motion vector estimation unit 102 are common. In some cases it is not necessary. In addition, the conversion function of this part needs to be appropriately changed depending on the combination of the signal input from the outside and the signal required in the subsequent stage.

  The motion vector estimation unit 102 estimates a motion vector from an image signal input as a YUV color space signal. A specific configuration is shown in FIG.

  The block dividing unit 201 divides the frame image (display image) of the input image signal into a plurality of regions (blocks) (FIG. 3). The reference block determination unit 202 sequentially sets the reference blocks so as to cover all regions of the frame image from the divided blocks. Here, the reference block means a block that is a starting point of the motion vector when the motion vector is to be estimated or detected.

The candidate block determination unit 203 determines a plurality of blocks (candidate blocks) within a predetermined search range, which are candidates for estimated motion vectors, each starting from the block designated as the reference block by the reference block determination unit 202.
Here, as one method for setting a predetermined search range, the entire frame can be set as the search range. In this case, all blocks included in the frame are searched (full search), and the reliability of the calculated motion vector is improved. On the contrary, only a part of the frame can be set as a search range (partial search). In this case, the calculation amount can be relatively reduced.

  The search range is set to a predetermined range over one or more frames before and / or after on the time axis with respect to the frame including the reference block (FIG. 3).

  The candidate block correlation value calculation unit 204 obtains correlation values between the reference block and individual candidate blocks for all candidate blocks. The correlation value is a value that is shown as the degree of approximation between blocks increases, such as the sum of absolute differences between pixels corresponding to the reference block and the candidate block, or the sum of squared differences between pixels corresponding to each of the reference blocks and candidate blocks. It may be a small one.

  In the calculation of the difference value, since each pixel in the case of a YUV image signal is composed of Y, Cb, and Cr signals (FIG. 4), the difference is obtained for all components (Y, Cb, and Cr). However, in this case, the calculation amount increases. In order to calculate at higher speed, the correlation value may be calculated by paying attention only to the luminance (Y) signal having a large influence on the image configuration. In this case, the calculation for Cb and Cr can be omitted, and the correlation value can be calculated at high speed.

  In the case shown in FIG. 4, the correlation value when considering all the pixels of YUV is 77 for the absolute difference total and 411 for the square difference. The correlation value when only the luminance signal (Y) is considered is 52 for the sum of absolute differences and 236 for the sum of squared differences.

  The estimated motion vector determination unit 205 sets a block having the highest correlation (low correlation value) as the movement destination block (estimation block) of the reference block from the correlation values of the candidate blocks calculated by the candidate block correlation value calculation unit 204. To do. Then, an estimated motion vector, an estimated motion vector correlation value, and a reference position are output. The estimated motion vector has a reference block as a starting point and an estimated block as an end point. The estimated motion vector correlation value is a correlation value between the reference block and the estimated block indicated by the estimated motion vector. The reference position is a position where the reference block is arranged on the frame.

  The motion vector estimation unit 102 sets all the blocks divided by the block division unit 201 as reference blocks in order, and calculates and outputs a correlation value between the estimated motion vector and another block for each reference block.

here,
The motion vector conversion unit 103 converts the input estimated motion vector based on the estimated motion vector calculated by the motion vector estimation unit 102, generates another new motion vector (conversion vector), and refers to the target block. Is.

  There are several ways to convert motion vectors. For example, regarding the calculation of the estimated motion vector in the motion vector estimation unit 102 as described above, a relatively large area in the image is the same, for example, when the entire screen moves uniformly or when a main subject constituting the screen moves. In the case of a simple motion state, there is a high possibility that the blocks around the reference block have the same motion vector as the reference block.

  In addition, regarding the movement of the image, if the uniformity of the direction in the movement of the entire screen or the subject is taken into consideration, the motion vector estimation unit 102 estimated from the reference block as a starting point among the blocks around the reference block. It is conceivable that a block located in substantially the same direction as the estimated motion vector is also likely to have a motion vector value similar to the reference block.

  Therefore, the motion vector conversion unit 103 mainly performs conversion regarding the direction of the vector with respect to the estimated motion vector value that is the output of the motion vector estimation unit 102.

  As specific conversion contents, for the input motion vector, a motion vector opposite in direction to the vector and approximately equal in size is calculated, or motion in the same direction as the vector and approximately equal in size. Such as calculating a vector. In short, a vector having the same magnitude in the direction substantially the same as or substantially opposite to the direction of the input motion vector is calculated.

  The content of the conversion in the motion vector conversion unit 103 is set to a motion vector value opposite in direction and equal in magnitude to the estimated motion vector of the motion vector estimation unit 102, and in the same direction as the vector and In the case of the motion vector values having the same magnitude, it is possible to control whether the motion detected in the reference block is estimated in the forward direction or in the backward direction. it can.

  Also, a vector obtained by adding the input motion vector and a vector having a predetermined magnitude in substantially the same direction is calculated, or a predetermined multiplication value is added to the input motion vector. There is also conversion such as calculating a vector obtained by multiplication.

  Here, in the motion vector conversion unit 103, assuming that a block specified by a conversion vector that is a converted vector is a target block, the motion vector conversion unit 103 includes a reference block that is a starting point of the conversion vector, and an end point of the conversion vector. In other words, the combination of target blocks to be generated is newly generated (combined).

  The content of the conversion in the vector conversion unit 103 is changed by adding the estimated motion vector of the motion vector estimation unit 102 and a vector having a predetermined magnitude in substantially the same direction or substantially the opposite direction, and changing the estimated motion vector In such a calculation, the displacement distance between the reference block and the target block can be controlled to increase or decrease independently with respect to the estimated motion vector detected in the reference block.

  In addition, when the content of the conversion in the vector conversion unit 103 is calculated by changing the magnitude by multiplying the estimated motion vector of the motion vector estimation unit 102 by a predetermined multiplication value, The displacement distance between the block and the target block can be controlled to increase or decrease dependently with respect to the estimated motion vector detected in the reference block.

  Such an estimated motion vector output from the motion vector estimation unit 102 and the motion vector conversion unit 103 and the converted motion vector have components in the horizontal direction and the vertical direction of the image signal, and these components are It is a finite value.

  Therefore, when the vector value converted by the motion vector conversion unit 103 is set for the position of the reference block (reference position) in the current frame, the direction of the vector value (substantially positive direction) starts from the reference block position. Alternatively, the position of a new block (target block) corresponding to the size and the opposite direction can be determined (FIG. 5). This target block is a block that has a high possibility of taking a motion vector similar to that of the reference block around the reference block.

  Further, the target block is not limited to one for one reference block. For example, other blocks through which a vector whose starting point is the position of the reference block on the screen and whose end point is the position of a certain target block can be similarly set as the target block (FIG. 6). In this case, the reference block may have motion vectors for a plurality of target blocks. Conversely, no target block can be assigned to all the reference blocks. In this case, there is no movement destination where the reference block moves within the screen.

  In the above description, a vector having a predetermined magnitude in substantially the same direction is added to the estimated motion vector. However, the characteristics of the input image signal, for example, the direction of movement of the entire screen and the magnitude of the movement are also included. Alternatively, the size of the vector to be added may be dynamically changed depending on the position where a clear boundary exists in the screen.

  Similarly, the estimated motion vector is multiplied by a predetermined multiplication value, but this may also be dynamically changed by using the characteristics of the input image signal. In that case, an appropriate motion vector can be detected from the input image signal.

  The correlation value calculation unit 104 calculates a correlation value calculated by the motion vector estimation unit 102 when the target block calculated from the motion vector conversion unit 103 is used as a reference block (FIG. 5). That is, the correlation value calculation unit 104 calculates the same correlation value that the motion vector estimation unit 102 calculates for the target block.

  Note that if the correlation value is calculated for all the blocks constituting the frame in the motion vector estimation unit 102, the correlation value calculation unit 104 does not have to repeat the same calculation performed by the motion vector estimation unit 102. By using (acquiring) the result, it is also possible to obtain a correlation value inherent to the target block.

  As another method, the correlation value calculation unit 104 may inherit / hold the correlation values for all the blocks calculated by the motion vector estimation unit 102 and select the correlation value of the target block therefrom. Is possible.

  The correlation values in the motion vector estimation unit 102 and the correlation value calculation unit 104 indicate the degree of correlation (degree of relevance) with respect to the respective motion vectors detected in the reference block and the target block. The smaller the value, the higher the degree of correlation with the block indicated by the motion vector, and the higher the certainty of the calculated motion vector value, that is, the higher the reliability.

  Next, the motion vector determination unit 105 takes in the estimated motion vector and correlation value calculated for the reference block, and the estimated motion vector and correlation value calculated for the target block (FIG. 5). The motion vector determination unit 105 determines the motion vector of the target block using these two correlation values.

  Specifically, the correlation value of the reference block calculated by the motion vector estimation unit 102 and the correlation value of the target block calculated by the correlation value calculation unit 104 are used, and both are compared or other threshold values are used. To determine the motion vector.

  For example, when the correlation value calculated by the motion vector estimation unit 102 (correlation value of the reference block) is smaller than the correlation value calculated by the correlation value calculation unit 104 (correlation value of the target lock), it is calculated by the motion vector estimation unit 102 The estimated motion vector of the determined reference block is determined as the motion vector of the target block.

  On the other hand, if the correlation value of the target block is smaller than the correlation value of the reference block, it is determined that the estimated motion vector of the target block is highly reliable, and the estimated motion vector is determined as a motion vector in the block.

When the correlation value of the reference block is equal to the correlation value of the target block, or when the difference between the two is within a certain range, one of the predetermined vectors is determined as the motion vector. Here, the certain range is, for example, (number of pixels constituting a block) × (range of values that each pixel can take) × allowable range (%)
It can also be determined by calculation as shown in.

  As another determination method, the correlation value of the reference block is compared with a predetermined threshold, and when the correlation value is smaller than the threshold, the estimated motion vector of the reference block is adopted, and when it is equal to or larger than the threshold, the target There is also a method of using an estimated motion vector of a block. According to this method, even when the correlation value of the reference block is smaller than the correlation value of the target block, the reliability of the estimated motion vector of the reference block is low when the correlation value of the reference block is generally large. As above, the estimated motion vector of the target block can be preferentially adopted.

  On the contrary, the correlation value of the target block is compared with a predetermined threshold value, and when the correlation value is larger than the threshold value, the estimated motion vector of the reference block is adopted. There is also a method that employs an estimated motion vector. According to this method, when the correlation value of the reference block is smaller than the correlation value of the target block and the correlation value of the target block is smaller than a certain threshold, the reliability of the estimated motion vector of the target block is determined. Therefore, the estimated motion vector can be preferentially adopted.

  In addition, when the correlation value of the target block is included in the range determined by the minimum value and the maximum value, the estimated motion vector of the target block can be adopted, and otherwise, the estimated motion vector of the reference block can be adopted. In this case, the estimated motion vector of the converted target block can be preferentially adopted as the range in which the correlation value of the target block can be evaluated as having high reliability.

  Furthermore, not only the estimated motion vector of the reference block and the estimated motion vector of the target block may be selected, but a new vector may be calculated from the plurality of vectors. For example, a new vector can be generated by adding each vector in proportion to the reliability of each correlation value of a plurality of vectors.

  In this way, by using the estimated motion vector as the target block calculated using not only the motion vector estimated for the reference block but also the motion vector of another block, a more accurate motion vector value can be effectively obtained. Can be detected.

  To utilize the above features of the present invention, to provide high-quality and high-efficiency processing in compression coding of an image signal subjected to motion compensation processing using a motion vector and generation of an interpolated image. Is possible.

  The motion vector estimation unit 102 and the correlation value calculation unit 104 receive the YUV signal and estimate the motion vector from the luminance signal (Y), but input an image signal in the HSV (hue, saturation, brightness) color space. It is also possible to estimate a motion vector using at least one of these elements. In this case, the motion vector can be estimated from changes in elements (hue, saturation, brightness, etc.) that are difficult to detect with luminance.

  Further, although the correlation value calculation unit 104 has described the method of using the correlation value calculated by the motion vector estimation unit 102, other methods are also possible. For example, the correlation value calculation unit 104 can perform the calculation again, and in this case, the correlation value can be calculated by a calculation method different from that of the motion vector estimation unit 102. When the correlation value calculation in the motion vector estimation unit 102 uses only the luminance signal, there is a method in which the correlation value calculation unit 104 calculates using all pixel elements. By doing so, since the correlation value is calculated with an index different from that of the motion vector estimation unit 102, it is possible to detect a motion vector with higher accuracy.

In the present invention, the reference block and the target block have been described. However, the reference block is the first reference block, the estimation block is the first estimation block, the target block is the second reference block, and the target block estimation block is the second. It is also possible to understand the present invention using the estimated motion vector of the first reference block as the first estimated motion vector and the estimated motion vector of the second reference block as the second estimated motion vector.
(Second Embodiment)
The difference between the present embodiment and the first embodiment is that a representative vector generation unit 706 is newly added and the internal processing of the motion vector determination unit 705 is partially different. Therefore, in the description of the present embodiment, only the points that differ from the first embodiment are described, and the other points are the same as in the first embodiment, and thus the description thereof is omitted.

  Vertical movement of characters at a specific horizontal position on the screen, such as horizontal movement of characters at a specific vertical position on the screen, such as subtitles and telops, or an end roll of a movie In the case of an image with direction movement, the reference block and other blocks in the same vertical position on the screen, or the reference block and other blocks in the same horizontal position on the screen, the reference block and its surrounding positions It can be easily estimated that there is a high possibility that the estimated motion vector is the same or similar for a block in

  Therefore, the representative vector generation unit 706 calculates a motion vector utilizing the above-described characteristics.

  The representative vector generation unit 706 receives the estimated motion vector calculated by the motion vector estimation unit 102 and its correlation value. Then, the motion vector values for a plurality of blocks around the reference block, a plurality of blocks at the same vertical position on the screen as the reference block, or a plurality of blocks at the same horizontal position on the screen as the reference block On the other hand, statistical processing represented by average value calculation is performed, and multiple blocks around the reference block, or multiple blocks in the same vertical position on the screen as the reference block, or on the reference block and the screen A representative vector value indicating a representative motion in a plurality of blocks at the same horizontal position is generated.

  Here, as a setting method of a plurality of blocks around the reference block, for example, a block adjacent to the reference block, a block included in a certain distance (range) from a certain point, or a certain distance from the reference block There are various methods such as a block or all blocks in the same frame as the reference block.

  A configuration example of the representative vector generation unit 706 is shown in FIG. The representative vector generation unit 706 receives the estimated motion vector correlation value, the estimated motion vector, and the reference position output from the motion vector estimation unit 102. Based on the input information, the switching unit 801 determines whether or not a representative vector should be generated at the reference position (reference block).

  For example, whether or not the representative vector can be generated can be determined based on whether or not the position of the reference block on the display screen corresponds to the screen edge. Since it is relatively difficult to calculate a motion vector at the screen edge, a motion vector can be obtained by using a representative vector as a motion vector.

  In the case of more advanced determination, it is possible to logically detect an image boundary from an input image signal and use a representative vector for calculating a motion vector near the boundary. Furthermore, it is representative of motion vector calculation by detecting the position where subtitles, telops, etc. are displayed, or the situation where a character string flows in the vertical direction (for example, end roll of a movie, etc.) from the content of the image signal. Vectors can be used. Since the switching unit 801 needs to determine whether representative vectors can be generated based on the estimated motion vectors of a plurality of blocks, the switching unit 801 includes a buffer (storage area) that stores a plurality of estimated motion vectors and their correlation values. You can leave it.

  When the representative vector is not generated, the input information is output as it is. In this case, the motion vector detection device 100 has the same contents as in the first embodiment.

  When the switching unit 801 generates a representative vector, the input information is sent to the representative vector calculation unit 802.

  The representative vector calculation unit 802 specifies a range in which a representative vector is generated from input information, for example, a block belonging to the same horizontal position, a block belonging to the same vertical position, or a block around the reference block. Then, a representative vector is generated from the estimated motion vector within the specified range by addition averaging or the like.

  The representative vector generation unit 706 outputs the representative motion vector by the above processing.

  Thus, the representative vector generated by the representative vector generation unit 706 represents the state of motion of the image in the reference block and the blocks around the reference block, and the movement of the reference block and the blocks around the reference block. Adaptation as a vector is possible.

  In addition, in the state where a part of the screen such as subtitles and telops moves uniformly, if the uniformity of the direction of the motion vector is taken into consideration, the blocks around the reference block start from the reference block as a representative vector. A block in the same direction as the representative vector generated by the generation unit 706 can be considered to have the same motion as this representative vector.

  The representative vector generated by the representative vector generation unit 706 is output to the motion vector conversion unit 103 and the motion vector determination unit 705. The motion vector conversion unit 103 converts the vector in the same manner as in the first embodiment, and the correlation value calculation unit 104 obtains a correlation value. The motion vector determination unit 805 determines and outputs a motion vector according to the representative vector and the converted motion vector.

  The motion vector determination unit 705 determines a motion vector based on the representative motion vector from the representative vector generation unit 706 and the correlation values from the motion vector conversion unit 103 and the correlation value calculation unit 104.

  Here, the difference from the first embodiment is that only the representative motion vector is input from the representative vector generation unit 706 and no correlation value is input, so that the motion vector determination unit 705 is the correlation from the correlation value calculation unit 104. The value is compared with a predetermined threshold value to determine which vector is used as the motion vector. The method for comparing the threshold value and the correlation value is the same as that in the first embodiment, and thus the description thereof is omitted.

  The configuration of other parts is the same as that of the first embodiment, and the description thereof is omitted.

  As described above, by adding the representative vector generation unit 706, it is possible to calculate a motion vector with higher accuracy when there is a feature in the motion of the screen.

  In any of the above embodiments, the input image signal has been described as being configured in units of frames, but the present invention is not limited to this. It is possible to apply the present invention in the same manner even if the input image signal is composed of field units. In that case, the predetermined search range set by the motion vector estimation unit is at least one or more fields before / after the field including the reference block.

  In any of the above embodiments, the motion vector detection device has been described. However, the present invention is not limited to this embodiment. The present invention can realize the above functions as an electronic circuit or an electric circuit. Further, it may be embodied as an integrated circuit (integrated IC or semiconductor chip) including the electronic circuit. Furthermore, the present invention may be realized as software operating on a CPU.

  When implemented as software, an example of the hardware configuration in which the software is executed is shown in FIG. For example, the input analog video signal is converted into an RGB digital signal by video A / D conversion and input to a CPU (DSP). The CPU (DSP) can detect a motion vector by executing software having a functional configuration as described in the above embodiment. At this time, previous and subsequent frame data and correlation value data that need to be temporarily stored in the motion vector estimation unit 102, the correlation value calculation unit 104, and the like are recorded in a memory connected to a CPU (DSP).

  Any of the above-described embodiments is illustrated for explaining the present invention, and is not limited to the above-described embodiment.

  The present invention can also be realized as a motion vector detection circuit and a motion vector detection device, and an integrated circuit or software incorporating these functions. MPEG or H262 encoding technology for digitally compressing an image, or a frame in a display device / Can be used for field interpolation technology.

The figure which shows the structural example of the motion vector detection apparatus in 1st Embodiment. The figure which shows the structural example of a motion vector estimation part The figure which shows the search range when estimating the motion vector The figure which shows the example of a difference at the time of calculating | requiring a correlation value between blocks The figure which shows the relationship between the target block calculated by a vector conversion part and a correlation value calculation part, and a target block The figure which shows the example (multiple object block) which calculates a conversion vector The figure which shows the structural example of the motion vector detection apparatus in 2nd Embodiment. An example of the configuration of the representative vector generator The figure which shows the hardware structural example for processing software of this invention

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Motion vector detection apparatus 101 YUV conversion part 102 Motion vector estimation part 103 Motion vector conversion part 104 Correlation value calculation part 105,705 Motion vector determination part 201 Block division part 202 Reference block determination part 203 Candidate block determination part 204 Candidate block correlation value Calculation unit 205 Estimated motion vector determination unit 706 Representative vector generation unit 801 Switching unit 802 Representative vector calculation unit

Claims (10)

In a motion vector detection circuit that detects a motion vector from an input image signal,
With respect to the first reference block is a display image of the partial area configuring the image signal, a first estimated motion vector estimating motion of the first reference block, pointed to the first estimated motion vector a motion vector estimation unit calculating a first correlation value indicating the magnitude of the correlation between the first estimation block and said first reference block, and
A motion vector conversion unit that calculates a conversion vector obtained by converting the size of the estimated motion vector in a predetermined method in a direction substantially the same as or substantially opposite to the first estimated motion vector ;
Correlation value for obtaining a second correlation value calculated when the motion vector estimation unit is applied to a second reference block indicated when the transformation vector is applied to the first reference block A calculation unit;
Based on the first correlation value and the second correlation value, and said first estimated motion vector, a second, which is calculated when applying the motion vector estimation unit in the second reference block A motion vector detection circuit comprising: an estimated motion vector; and a motion vector determination unit that determines a motion vector of the second reference block from the estimated motion vector . The motion vector conversion unit calculates a conversion vector having substantially the same magnitude as the first estimated motion vector in a direction substantially the same as or substantially opposite to the first estimated motion vector ;
The motion vector detection circuit according to claim 1 . The motion vector conversion unit calculates a conversion vector obtained by adding a predetermined magnitude to the first estimated motion vector in a direction substantially the same as or substantially opposite to the estimated motion vector ;
The motion vector detection circuit according to claim 1 . The motion vector conversion unit calculates a conversion vector having a magnitude obtained by multiplying a magnitude of the first estimated motion vector by a predetermined value in a direction substantially the same as or substantially opposite to the first estimated motion vector. ,
The motion vector detection circuit according to claim 1 . The motion vector conversion unit calculates a plurality of conversion vectors for one first estimated motion vector .
The motion vector detection circuit according to claim 1 . The motion vector determination unit determines and outputs the first estimated motion vector as a motion vector of the second reference block when the first correlation value is smaller than or equal to the second correlation value ;
The motion vector detection circuit according to claim 1 . The motion vector determination unit determines the first estimated motion vector as a motion vector of the second reference block when the first correlation value is smaller than a threshold ;
The motion vector detection circuit according to claim 1 . The motion vector determination unit determines the first estimated motion vector as a motion vector of the second reference block when the second correlation value is greater than a threshold ;
The motion vector detection circuit according to claim 1 . Integrated circuit including the motion vector detecting circuit according to any one of claims 1 to 8. Motion vector detecting device including a motion vector detecting circuit according to any one of claims 1 to 8.

Images