JP2000333184A - Motion vector detection method, device therefor and data recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce the calculation frequency without deteriorating the encoding efficiency by searching a position on a reference screen of an estimated block under a fixed condition, calculating an error of image data between a block corresponding to the searching center of a motion vector(MV) and a block to be processed and comparing the motion compensation error corresponding to a nearby block with the motion error corresponding to searching center of the motion vector. SOLUTION: A discontinuation deciding unit u5a outputs the deciding data Dc showing the discontinuation of the searching processing of a 2nd layer if the motion compensation data D1p set to the motion searching center are smaller than the reference error data Drp set to a nearby block. If the data D1p is larger than the data Drp, the data Dc showing the non-discontinuation of the searching processing of the 2nd layer is outputted. When the deciding unit u5a decides the discontinuation of the searching processing of the 2nd layer, a fact that no arithmetic processing is necessary is notified to an MV generator u1, etc.

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, a motion vector detecting device, and a data storage medium. The present invention relates to a motion vector detection process for quickly and accurately generating a motion vector used in a process of efficiently encoding for each block composed of a predetermined number of pixels.

[0002]

2. Description of the Related Art In recent years, the multimedia era in which voices, images, and other data are handled in an integrated manner has been entered, and conventional information media, that is, means for transmitting information such as newspapers, magazines, televisions, radios, and telephones to humans. Has been taken up as an object of multimedia. Generally, multimedia means not only characters but also graphics, sounds, and especially images, etc. are simultaneously associated with each other. To make the above-mentioned conventional information media the target of multimedia, the information must be expressed in digital form. Is an essential condition.

However, when the amount of information handled by each of the above information media is estimated as a digital information amount, the amount of information per character is 1 to 2 bytes in the case of characters, whereas the amount of information per character is 1 to 2 bytes in the case of voice. 64Kbits (telephone quality), plus 100Mbits per second for video
(Current television broadcast quality) or more is required, and in most of the above-mentioned information media, it is not realistic to handle the vast amount of information in digital form as it is. For example, a videophone is 64Kbps to 1.5Mbp.
service integrated digital network (ISD)
N: Integrated Services Digital Network), but it is impossible to send the video information of a TV camera by ISDN as it is.

Therefore, what is needed is an information compression technique. For example, in the case of a videophone, ITU-T
(International Telecommunication Union Telecommunication Standardization Sector) 261 and H.E. H.263 video compression technology is used. According to the information compression technology of the MPEG1 standard, a normal music CD (compact disc) is used.
It is also possible to record image information together with audio information.

Here, MPEG (Moving Picture Exper)
ts Group) is an international standard relating to a technique for compressing moving image data (image signals of moving images). It is a standard. MPE
Transmission speed for G1 standard is mainly about 1.5Mbps
Since it is limited to s, the moving image data is compressed to 2 to 15 Mbps in MPEG2 standardized to meet the demand for higher image quality.

Further, at present, MPEG1, MPEG2
And the working group that has promoted standardization (ISO / IEC JTC1 / SC2
9 / WG11), a moving image data compression technology that enables encoding processing and signal operation on an object basis and realizes new functions required in the multimedia age is being standardized as MPEG4. At first, MPEG4 aimed to standardize a low bit rate encoding method.
The standardization target has been extended to a more general-purpose encoding process of a high bit rate corresponding to an interlaced image.

In the above-described encoding process, a large amount of data is compressed (bit rate saving) by using a process of encoding an image signal using pixel value correlation between screens, that is, a so-called inter-frame motion compensation encoding process. Has been realized. Here, in the inter-screen motion compensation coding processing, an image space (macroblock) composed of 16 × 16 pixels or 8 × 16 pixels is used.
In units of an image space (block) composed of eight pixels, an image signal of a target block to be encoded (hereinafter, also referred to as a target block) is converted into an image signal corresponding to a screen on which the encoding process has been performed. (Image signal of the reference screen), position information indicating the position of the region (prediction block) in which the array pattern of the pixel values is most similar to the target block on the reference screen is determined as the motion vector corresponding to the target block. Is detected. This motion vector is encoded together with a difference signal (motion compensation error) between the image signal corresponding to the target block and the image signal corresponding to the prediction block (reference block).

FIG. 11 is a block diagram of a typical conventional inter-picture motion compensation coding apparatus. For the sake of simplicity, in the following description, a macro block composed of 16 × 16 pixels and a block composed of 8 × 8 pixels are referred to as blocks without distinction. This inter-screen motion compensation coding device (hereinafter, referred to as
Abbreviated as encoding device. And 1000, an image signal obtained by dividing an image signal corresponding to a predetermined object so as to correspond to the above-described block, is sequentially received for each block, and an inter-screen motion compensation encoding process is performed on the image signal corresponding to each block. Is applied.

That is, the encoding apparatus 1000 includes an image signal St corresponding to a target block to be encoded and an image signal (motion compensation) corresponding to a prediction block obtained from an image signal corresponding to a reference screen. Data) a subtractor d1 for calculating a difference value Dt from Pt, and a difference value Dt obtained by the subtractor d1 is converted to a DCT (discrete cosine transform).
An orthogonal transformer d2 for converting the frequency component Tt into a frequency component Tt by an orthogonal transform process, a quantizer d3 for quantizing the frequency component Tt and outputting a quantized value Qt, and a coded data for coding the quantized value Qt. And a variable length encoder d4 for outputting Et.

[0010] The encoding apparatus 1000 also includes an inverse quantizer d5 for inversely quantizing the quantized value Qt and restoring the frequency component IQt, and an inverse DCT for the restored frequency component IQt.
Image data (pixel value) in the spatial region by processing such as T processing
An inverse orthogonal transformer d6 for converting to ITt, an adder d7 for adding the restored image data ITt and the motion compensation data Pt, and an output Rt of the adder d7 is referred to at the time of encoding processing for a subsequent screen. And a frame memory d8 for temporarily storing the data as the data for use.

[0011] Further, the encoding apparatus 1000 generates an image signal St corresponding to the block to be processed, for each region on the reference screen having the same size as the block to be processed.
Is compared with the image data Ct of the reference screen stored in the frame memory d8, and information (motion vector) MV indicating the position of the area (prediction block) on the reference screen where the difference between the two is minimized is detected. And an address signal Ad for the frame memory d8 corresponding to the detected motion vector MV. Based on the address signal Ad, the image data Mt of the area (prediction block) specified by the pixel position on the reference screen is generated. A motion compensator d10 that acquires and outputs this image data as motion compensation data Pt.

In the above-described conventional encoding apparatus 1000,
Instead of directly coding the image signal St, the compression efficiency is improved by coding the output of the subtractor d1 having a small energy. It is known that the compression efficiency is improved as the energy of the output of the subtractor d1 (the amplitude of the output signal) is smaller.
9, it is important to detect a motion vector quickly and accurately. By detecting such an accurate motion vector, the energy of the output of the subtractor d1 is reduced.

In the figure, i20 is an input terminal for an image signal (image data) St, and o20 is an output terminal for coded data Et. In the encoding device 1000 shown in FIG. 11, a motion vector detecting device 200 for detecting a motion vector is configured by a frame memory d8 and a motion detector d9. Hereinafter, the motion vector detecting device 200
Will be described in detail.

FIG. 12 is a block diagram showing a configuration of a conventional typical motion vector detecting device. The motion vector detecting device 200 receives a frame memory d8 in which the image data Ct of the reference screen is stored, and information Sc indicating a motion vector search center from the outside, and receives a candidate motion indicating a plurality of pixel positions near the search center. A motion vector generator u1a that outputs a vector MVc, a pixel value (image data) Ct of a prediction block in a reference screen indicated by a candidate motion vector MVc, and an input image signal (image data) St of a block to be processed. Error (motion compensation error) D
an error calculator u3a for calculating p, and the error calculator u3a
Monitor the error Dp, which is the output of
A minimum error selector u4a that selects a candidate with the minimum error from the candidate motion vector MVc that is the output of 1a, and outputs the selected candidate motion vector as a motion vector MVs corresponding to the block to be processed. I have. Here, the motion detector d9 includes the motion vector generator u
1a, an error calculator u3a, and a minimum error selector u4a.

In the figure, i1 is a first input terminal of the motion vector detecting device 200 to which the image signal (image data) St is input, and i2 is a motion vector detecting device to which the motion vector search center information Sc is input. A second input terminal o1 of the device 200 is an output terminal of the detection device 200 from which the motion vector MVs is output. The motion vector search center information Sc indicates a position at which a search process for detecting a motion vector by comparing the image signal Ct of the reference screen stored in the frame memory d8 with the input image signal St is started. The start position of this search processing is indicated by a position on the reference screen corresponding to the same spatial position as the target block on the processing target screen, or a detected motion vector of a neighboring block located near the target block. The position on the reference screen is often used. For example,
As a neighboring block located near the target block, a block located on the left of the target block is used.

Next, the operation will be described. When an image signal St corresponding to a target block to be encoded is input to the inter-screen motion compensation encoding apparatus 1000, the image signal St of the target block is input to a subtractor d1.
, And a difference value Dt between the image signal (motion compensation data) Pt of the corresponding prediction block. This difference value Dt
Is DCT (discrete cosine transform) by the orthogonal transformer d2
Is converted into a frequency component Tt by orthogonal transformation processing such as.
This frequency component Tt is quantized by a quantizer d3,
The corresponding quantized value Qt is output. Then, in the variable length encoder d4, a variable length encoding process for converting the quantized value Qt into a variable length code is performed, and the encoded data Et corresponding to the block to be processed is output from the encoder d4. Is output.

On the other hand, at this time, the quantized value Qt is converted into a frequency component IQt by an inverse quantization process in an inverse quantizer d5.
The frequency component IQt is further converted into image data (pixel value) ITt in the spatial domain by the inverse orthogonal transformer d6. Then, the image data ITt is added to the motion compensation data (image signal corresponding to the prediction block) Pt by the adder d7 to generate reproduction data Rt corresponding to the block to be processed. This playback data Rt
Is temporarily stored in the frame memory d8 as reference data for reference when encoding an image signal corresponding to a subsequent screen following the processing target screen.

In the motion detector d9, the image signal Ct corresponding to the reference screen stored in the frame memory d8 is compared with the image signal St corresponding to the block to be processed. The position information of the predetermined area (prediction block) having the same size as that of the block to be processed is derived as the motion vector MV corresponding to the block to be processed. Then, in the motion compensator d10, an address Ad corresponding to the position on the reference screen indicated by the motion vector MV is generated. , Are output as motion compensation data Pt corresponding to the block to be processed.

Next, the operation of the motion vector detecting device 200 including the motion detector d9 and the frame memory d8 will be described in detail. The motion vector detecting device 20
The motion vector generator u1a at 0 outputs a motion vector indicating a plurality of pixel positions near the search center as a candidate motion vector MVc based on the vector search center information Sc supplied from outside the device. Then
In the frame memory d8, the candidate motion vector MV
An image signal (motion compensation data) Ct corresponding to the block on the reference screen indicated by c is output. The error calculator u3a calculates a difference signal (image error data) Dp between the motion compensation data Ct corresponding to each candidate motion vector and the image signal of the target block. The minimum error selector u4a monitors the image error data Dp, selects a candidate motion vector MVc with the minimum error data Dp from a plurality of candidate motion vectors MVc, and selects the selected candidate motion vector. The motion vector (selected motion vector) MVs corresponding to the processing block is output.

Although the motion vector detecting device 200 has a simple structure, it calculates a motion compensation error in proportion to the search range of the motion vector, that is, the number of candidate motion vectors generated by the motion vector generator u1a. However, there is a disadvantage that the amount of calculation when performing the operation is increased. In other words, for an image including an object having a large motion, searching for a motion vector over a wide range in the image space leads to an improvement in coding efficiency. It is important to reduce the error calculation with the largest calculation load among the calculation processes in the motion vector detection device. Therefore, a hierarchical motion vector detection method is often used as a method for reducing the number of error calculation operations in the above-described motion vector detection device.

FIG. 13 is a diagram for explaining a conventional motion vector detection method (three-step method) which is an example of such a hierarchical motion vector detection method. In the three-step method, in the processing of the n-th layer, the search is performed at eight neighboring positions around the designated position indicated by the motion vector detected in the processing of the (n-1) -th layer. The range is set to 1 of the search range in the processing of the (n-1) th hierarchy.
/ 3 and recursively search.
In addition, the above eight neighboring positions are in the upward direction of the designated position,
These are four neighboring positions located in the downward, left, and right directions, and four neighboring positions located in the upper right, lower right, upper left, and lower left directions of the designated position.

In FIG. 13, the circles represent the respective pixels constituting the reference screen, and the circles with the number 1 written therein are the pixels (1) to be searched in the first hierarchical processing. (First search target pixel), which is a representative position of a candidate prediction block (motion compensation block) indicated by a candidate motion vector. The reference screen shown in FIG. 13 shows nine first search target pixels including the pixel serving as the search center.

First, in the first hierarchical motion detection process, image error data corresponding to a specific block specified by the position of the pixel P0 and its surrounding pixels P1a to P1h (ie, the pixel P0) is used as a search center. An error in the image data between the specific block and the block to be processed) is detected. Here, the peripheral pixel P1b has the smallest image error data. Accordingly, the second hierarchical motion detection process is performed with this pixel P1b as the search center. The peripheral pixels P1a to P1h
Are located on a straight line parallel to the vertical direction, the horizontal direction, and the oblique direction at an angle of ± 45 degrees with respect to the horizontal direction, including the search center, from the pixels P0 to P9 serving as the search center.
They are located at pixel intervals.

Next, in the second hierarchical motion detection processing, the above-described pixel P1
Image error data corresponding to a specific block specified by the positions of the pixels P2a to P2h around b (that is, an error in the image data between the specific block and the block to be processed) is detected. Here, the peripheral pixel P2e is
This is the smallest image error data. Accordingly, the motion detection processing of the third hierarchical level is performed for this pixel P2e.
The search is mainly performed. The peripheral pixel P2a
To P2h on the straight line parallel to the vertical direction, the horizontal direction, and the oblique direction forming an angle of ± 45 degrees with respect to the horizontal direction, including the search center in the second-layer motion detection process. It is located at a distance of three pixels from the pixel P1b serving as the search center.

Finally, in the third-layer motion detection process, the peripheral pixel P2e is set as the search center, and the pixel P2
Image error data corresponding to a specific block specified by the positions of the pixels P3a to P3h around e (that is, an error in image data between the specific block and the block to be processed) is detected. Here, the peripheral pixel P3a is
This is the smallest image error data. Therefore, the motion vector indicating the pixel P3a (the hatched circle) is the motion vector corresponding to the block to be processed detected by the three-step method. The peripheral pixels P3a to P3h are parallel to the vertical direction, the horizontal direction, and the oblique direction forming an angle of ± 45 degrees with respect to the horizontal direction, including the search center in the third hierarchical motion detection process. On a straight line, the pixel P2e that is the search center is located at a distance of one pixel.

In general, the three-step method simply satisfies the condition that the image data is almost satisfied, that is, the condition that the pixel closer to the search position where the image error data is minimum has the corresponding image error data smaller. The image error data is directly calculated for all the pixel positions in the search range (that is, all the pixels in the reference screen), and the motion is detected with almost the same accuracy as the method for detecting the motion vector with the minimum image error data. Vectors can be detected.

Further, in the motion detection processing by the three-step method shown in FIG. 13, the number of times of calculation of the image error data (hereinafter simply referred to as an error) is nine in the processing of the first hierarchy and eight in the processing of the second hierarchy. This is a total of 25 times, ie, 8 times in the processing of the third hierarchy, and it is clear that the number of times of calculating the error can be greatly reduced as compared with a method of directly calculating the error for all the pixel positions in the search range. is there.

FIG. 14 is a diagram for explaining another conventional motion vector detection method. In this motion vector detection method, a search process for a pixel position at which an error (image error data) is minimized is performed with half-pixel accuracy.

In the above-described 1/2 pixel precision search processing, an interpolated pixel having a virtual pixel value is defined between two adjacent pixels on the reference screen, and on the reference screen including the interpolated pixel, A search for a motion vector indicating a pixel position where the image error data is minimized is performed. On the basis of the motion vector of 1/2 pixel accuracy obtained by this search processing,
High-precision motion compensation for pixels or less is performed. Note that the pixel value of the interpolation pixel is obtained by performing an interpolation operation on the pixel values of the pixels located on both sides of the interpolation pixel.

Such a search process with half-pixel accuracy is as follows.
It is used to improve the coding efficiency by reducing the energy (amplitude level) of the output signal of the differentiator d1 in the coding device shown in FIG.

In the half-pixel precision search processing, when reading out the pixel value of the reference screen from the frame memory, a virtual pixel value of an interpolated pixel located between two adjacent pixels is generated by an interpolation operation. New processing is required. For this reason, in the search processing of the first layer, motion detection processing with integer pixel accuracy is performed on pixels indicated by solid circles in FIG. 14, and in the search processing of the second layer, the first layer search processing is performed. The pixel P′0 (see FIG. 14) detected at the search center is used as a search center, and a pixel to be searched is located around and adjacent to the pixel P′0 (pixels P′1 to P′1 in FIG. 14).
Only at P′8), a pixel position where the error is minimized is searched. As a result, a motion vector with 1/2 pixel accuracy is detected. The feature of these conventional hierarchical motion vector search processes is that while gradually narrowing the search range and increasing the search density, the motion vector that always has the minimum error within the search range corresponding to each stage is selected. That is.

FIG. 15 is a block diagram for explaining a conventional hierarchical motion vector detecting device. The hierarchical motion vector detecting device 201 detects a motion vector MV1s by a first hierarchical search process based on position information Sc (hereinafter, also referred to as motion vector search center information) indicating a motion vector search center input from the outside. And a second-layer motion detecting unit that detects a motion vector MV2s by a second-layer search process based on the motion vector MV1s obtained by the first-layer search process. I have.

Here, the above-mentioned first-layer motion detecting section is constructed as shown in FIG.
MV generator u1a that generates a plurality of first candidate motion vectors MV1c based on external motion vector search center information Sc, similarly to motion vector detection device 200 shown in FIG.
, A frame memory u2 storing image data corresponding to the reference screen, and the first candidate motion vector MV1c.
Motion compensation data (image data of a candidate prediction block) C1t read from the frame memory u2 based on
And an error calculator u3a for calculating error data D1p between the input target block image data St and a plurality of candidate motion vectors MV1 based on the error data D1p.
c), a minimum error selector u4a for selecting a candidate motion vector having the minimum error data as the first-layer motion vector MV1s.

The second-layer motion detecting section includes an MV generator u1b for generating a plurality of second candidate motion vectors MV2c around a pixel position indicated by the first-layer motion vector MV1s as a search center, and a reference screen corresponding to the reference screen. And a motion compensation data (image data of a candidate prediction block) C2t read from the frame memory u2 based on the second candidate motion vector MV2c. An error calculator u3 for calculating error data D2p from the image data St of the target block
b, and a minimum error selector u4b that selects a candidate motion vector with the minimum error data from among the plurality of candidate motion vectors MV2c as the second hierarchical motion vector MV2s based on the error data D2p. I have. The error calculator u3b of the second layer motion detecting unit includes the first
The minimum error data D1pm is supplied from the minimum error selector u4a of the hierarchical motion detection unit. For this reason, it is not necessary for the error calculator u3b to obtain error data corresponding to the search center in the second hierarchical motion detection processing.

In the motion vector detecting device 201, the frame memory u2 includes the first hierarchical motion detecting unit and the second hierarchical motion detecting unit.
It is shared by the hierarchical motion detector. FIG.
5, i1 is a first input terminal to which the image signal St of the target block is input, i2 is a second input terminal to which the motion vector search center information Sc is input, and o1 is a motion vector MV2s of the target block. Output terminal to be output. Note that the frame memory u2 shown in FIG.
And the frame memory d8 shown in FIG.

Next, the operation will be described. The first-layer motion detection unit of the motion vector detection device 201 shown in FIG. 15 performs exactly the same search processing as the motion vector detection device 200 shown in FIG. It is generated as a vector MV1s.

When the first hierarchical motion vector MV1s is input to the second hierarchical motion detecting section of the motion vector detecting device 201, the second hierarchical motion detecting section displays the first hierarchical motion vector MV1s on the reference screen indicated by the first hierarchical motion vector MV1s. Is set as the search center, a plurality of second candidate motion vectors MV2c corresponding to the peripheral pixels located near the search center are generated.

Further, the error calculator u3b receives image data (motion compensation data) C2t corresponding to the candidate prediction block on the reference screen indicated by the second candidate motion vector, which is read from the frame memory u2. Then, error data D2p between the motion compensation data C2t and the image data St of the target block is calculated.

The minimum error selector u4b includes an MV generator u
In addition to the plurality of candidate motion vectors MV2c generated in 1b, error data D2p corresponding to each candidate motion vector MV2c is input, and the candidate motion vector MV2c with the minimum error data is selected as the second-layer motion vector MV2s. Is output.

The error data D1pm corresponding to the search center of the motion vector generator u1b (ie, the motion vector selected by the minimum error selector u4a) has already been calculated by the error calculator u3a, and has been calculated by the minimum error selector u4a. Since the minimum error data D1pm is notified from the minimum error selector u4a to the error calculator u3b, the error calculator u3b calculates the motion vector indicating the search center in the second hierarchical motion search process. In addition, the calculation for obtaining the corresponding error data can be omitted.

FIG. 15 shows a motion vector detecting device 201 which detects a motion vector corresponding to a target block by a two-layer search process.
FIG.
It can be easily analogized that the motion vector detection device 200 shown in FIG. 2 can be obtained by further adding one set of an error calculator, a minimum error selector, and an MV generator.

The conventional motion vector detection method is not limited to the method of performing the hierarchical search process described with reference to FIGS. 13 and 14, and the motion vector search center is used as a search start position for each pixel position. In some cases, a search direction is determined to search for a pixel position on a reference screen where the motion compensation error becomes smaller. FIG. 16 is a diagram for explaining a motion vector detection method for sequentially moving the search position from the vector search center by obtaining a pixel position at which the motion compensation error becomes smaller, and the arrangement of pixels on the reference screen. Is shown. In the drawing, Ps0 and Psa to Psi are pixels on the reference screen.

This motion vector detection method is based on One at at
ime, and the processing flow of this motion vector detection method will be specifically described below. However, here, the error (motion compensation error) of the image data between the block on the reference screen specified by the pixels Ps0 and Psa to Psi shown in FIG. 16 and the block to be processed on the screen to be processed.
For SADs0, SADsa to SADsi, the following equation (1)
It is assumed that (3) holds. SADs0, SADsa> SADsb (1) SADsb, SADsd> SADsc (2) SADsc, SADsf to SADsi> SADse (3) In this motion vector detection method, first, in the first step, On the reference screen, a search is performed for the pixels Psa and Psb located on the left and right of the pixel Ps0 serving as the search center.

In this case, the motion compensation error SADsb corresponding to the pixel Psb on the right side of the search center pixel Ps0 is the motion compensation error SAD corresponding to the search center pixel Ps0 and the pixel Psa on the left side thereof.
It is smaller than ADs0 and SADsa (see equation (1)). Therefore, in the subsequent second step, the pixel P
A search is performed for pixels Ps0 and Psc located on the left and right of sb.

In the third step, the motion compensation error SADsc corresponding to the pixel Psc is calculated by using the pixel Ps0 and the pixel Psb.
Is smaller than the motion compensation errors SADs0 and SADsb corresponding to the pixel P (see equations (1) and (2)).
A search is performed for pixels Psb and Psd located on the left and right of sc.

In the fourth step, the motion compensation error SADsc corresponding to the pixel Psc is smaller than the motion compensation errors SADsb and SADsd corresponding to the pixels Psb and Psd (see equation (2)). A search is performed for pixels Pse and Psf located above and below pixel Psc.

In the fifth step, the motion compensation error SADse corresponding to the pixel Pse is smaller than the motion compensation errors SADsc and SADsf corresponding to the pixels Psc and Psf (see equation (3)). A search is performed for pixels Psg and Psh located to the left and right of pixel Pse.

In the sixth step, the motion compensation error SADse corresponding to the pixel Pse is smaller than the motion compensation errors SADsg and SADsh corresponding to the pixels Psg and Psh (see equation (3)). A search is performed for the pixels Psi and Psc located above and below the pixel Pse. In this case, since the motion compensation error SADse corresponding to the pixel Pse is smaller than the motion compensation errors SADsi and SADsc corresponding to the pixel Psi and the pixel Psc (see Equation (3)), the position of the pixel Pse Is determined as the motion vector for the block to be processed. The numbers in the circles indicating the pixels in FIG. 16 indicate the pixels newly added to the search target in the first to sixth steps. For example, the number in the circle indicating the pixel Psc is 2, and this number 2 indicates that this pixel Psc is newly added to the search target in the second step.

[0049]

As described above,
In the conventional motion vector detection processing, the number of times of calculating image error data has been greatly reduced by a hierarchical motion vector search or an efficient motion vector search such as a one-at-time search. As the demands for energy saving and energy saving have become more and more strict, it has become necessary to reduce the number of times of calculating error data in the motion vector detection processing.

It is needless to say that the coding efficiency of an image signal (improvement of image quality at a constant bit rate) is important for applications such as the above-described energy-saving type small-sized image display device. There is a problem that it is required to reduce the number of calculations in a process with a large amount of calculation such as detection.

The present invention has been made in view of the above-described problems, and it is possible to reduce the number of calculations in a motion vector detection process requiring a large amount of calculation without substantially deteriorating the coding efficiency of an image signal. A motion vector detection method and a motion vector detection device capable of suppressing power consumption by this, and a data storage medium storing a program for causing a computer to perform a process of detecting a motion vector by the motion vector detection method. The purpose is to get.

[0052]

A motion vector detecting method according to the present invention (claim 1) performs processing for each block of a predetermined number of pixels, which is a unit for processing image data corresponding to a screen to be processed. Motion vector detection for searching for a position on a reference screen of a prediction block corresponding to a target block to be processed under a predetermined condition, and detecting information indicating the position of the prediction block as a motion vector corresponding to the block to be processed. A motion vector detection method including a process, wherein a block corresponding to a motion vector search center that is a position where a search for a position of the prediction block is started, and image data between the block to be processed to be processed. A calculation process of calculating an error as a motion compensation error corresponding to the motion vector search center; And a comparison process of comparing a motion compensation error corresponding to a neighboring block, which is an error of image data between a lock and a corresponding prediction block, with a motion compensation error corresponding to the motion vector search center. ,
According to the comparison result between the two motion compensation errors, the position information indicating the motion vector search center is output as a motion vector corresponding to the block to be processed, or the motion vector detection process is performed to search the motion vector search center. Switching is performed as to whether a motion vector obtained as a start position is output as a motion vector corresponding to a block to be processed.

According to a second aspect of the present invention, in the motion vector detecting method according to the first aspect, when the motion compensation error corresponding to the motion vector search center is smaller than the motion compensation error corresponding to the neighboring block. The position information indicating the vector search center is output as it is as a motion vector corresponding to the block to be processed.

According to a third aspect of the present invention, in the motion vector detecting method according to the first aspect, a motion compensation error corresponding to the motion vector search center is smaller than a motion compensation error corresponding to the neighboring block, and When the motion compensation error corresponding to the vector search center is less than a predetermined value,
The position information indicating the motion vector search center is output as it is as a motion vector corresponding to the block to be processed.

According to a fourth aspect of the present invention, in the motion vector detecting method according to the third aspect, when a motion detection command signal is input from outside, a motion compensation error corresponding to the neighboring block and the motion vector search center are obtained. Irrespective of the magnitude relation of the motion compensation error corresponding to the motion vector detection process, the motion vector obtained by performing the motion vector detection process with the motion vector search center as the search start position is output as the motion vector corresponding to the block to be processed. is there.

According to a fifth aspect of the present invention, in the motion vector detecting method according to the first aspect, the difference between the motion compensation error corresponding to the neighboring block and the motion compensation error corresponding to the motion vector search center is large. Is smaller than a predetermined value, the position information indicating the motion vector search center is output as a motion vector corresponding to the block to be processed.

In the motion vector detecting method according to the present invention (claim 6), a motion vector to be processed is processed for each block of a predetermined number of pixels, which is a unit for processing image data corresponding to a screen to be processed. A motion vector detecting method for searching a position of a prediction block corresponding to a processing block on a reference screen under a predetermined condition, and detecting information indicating the position of the prediction block as a motion vector corresponding to the block to be processed. , The first set on the reference screen
Is searched for in the search range, and a block on the reference screen specified by the first candidate motion vector corresponding to the block to be processed and the block to be processed on the screen to be processed are searched. Calculating a first motion compensation error, which is an error in the image data of the first candidate motion vector, and performing the prediction in the second search range on the reference screen located near the pixel position indicated by the first candidate motion vector. This is the error of the image data between the block on the reference screen specified by the second candidate motion vector corresponding to the block to be processed and the block to be processed on the screen to be processed. Calculating a second motion compensation error; and calculating a position of the prediction block in a third search range on the reference screen set based on the first candidate motion vector and the second candidate motion vector. A third search which is an error of image data between the block on the reference screen specified by the third candidate motion vector corresponding to the block to be processed and the block to be processed on the screen to be processed. The processing for calculating the motion compensation error is compared with the first, second, and third motion compensation errors, and the first, second, and third candidate motion vectors are compared according to the comparison result. And selecting one as a motion vector corresponding to the block to be processed.

According to a seventh aspect of the present invention, in the motion vector detecting method according to the sixth aspect, the third search range is set to an area other than the first and second search ranges on the reference screen. The search for the position of the prediction block in the third search range is performed without referring to the first and second motion compensation errors.

According to the present invention (claim 8), in the motion vector detecting method according to claim 6, in the first search range, a search process for each pixel is performed as a process of searching for the position of the prediction block. In the third search range, the process of searching for the position of the prediction block is performed with an accuracy of half a pixel or more by interpolating pixel values corresponding to virtual pixels between pixels on the reference screen.

According to a ninth aspect of the present invention, in the motion vector detecting method according to the sixth aspect, in the second search range, the position of the prediction block is set only in the vertical or horizontal direction on the reference screen. In the third search range, the position of the prediction block is searched for in a direction other than the search direction in the second search range.

According to the present invention (invention 10), the motion vector estimating apparatus includes a unit for processing image data corresponding to a screen to be processed. For the corresponding prediction block,
A motion vector detecting device that performs a motion vector detection process of searching for a position on a reference screen under a predetermined condition and detecting information indicating the position of the predicted block as a motion vector corresponding to the block to be processed. The error of the image data between the block corresponding to the motion vector search center, which is the position where the search for the position of the block is started, and the processing target block to be processed is calculated by the motion compensation corresponding to the motion vector search center. Error calculating means for calculating an error, a neighboring block located in the vicinity of the block to be processed, and a motion compensation error corresponding to a neighboring block which is an error of image data between the corresponding predicted block and the motion compensation error. Comparing means for comparing with a motion compensation error corresponding to the motion vector search center; and Discriminating means for discriminating whether or not to perform motion vector detection process on the serial target processing block,
Only when it is determined that the motion vector detection processing should be performed by the determination means, a motion detection means for detecting the motion vector by performing the motion vector detection processing with the motion vector search center as a search start position; and When it is determined that the motion vector detection process should be performed, the motion vector detected by the motion detection unit should be output as a motion vector corresponding to the block to be processed, and the motion vector detection process should not be performed by the determination unit. When it is determined that there is no motion vector, the selection unit that outputs the position information indicating the motion vector search center as it is as a motion vector corresponding to the block to be processed, and a motion compensation error corresponding to the motion vector output from the selection unit, Data for determining whether or not to perform a motion vector detection process on another block; Those having a memory for storing temporarily Te.

According to the motion vector detecting device of the present invention (claim 11), for each block of a predetermined number of pixels, which is a unit for processing image data corresponding to a screen to be processed,
A motion vector for searching for a position on a reference screen of a prediction block corresponding to a target block to be processed under a predetermined condition, and detecting information indicating the position of the prediction block as a motion vector corresponding to the target block. A detection device that searches for a position of the prediction block in a first search range set on the reference screen, and detects a position of the prediction block on a reference screen specified by a first candidate motion vector corresponding to the block to be processed. Motion detecting means for calculating a first motion compensation error, which is an error of image data between the block of interest and the block to be processed on the screen to be processed, and a pixel indicated by the first candidate motion vector In the second search range on the reference screen located near the position, the position of the prediction block is searched for, and a second candidate motion vector corresponding to the block to be processed is used. And a block on the reference screen that is constant,
The second, which is the error of the image data with the block to be processed
A second motion detecting means for calculating a motion compensation error of the prediction block, and a prediction block in a third search range on the reference screen set based on the first candidate motion vector and the second candidate motion vector. , And a third motion compensation which is an error of the image data between the block on the reference screen specified by the third candidate motion vector corresponding to the block to be processed and the block to be processed. The third motion detecting means for calculating the error is compared with the first, second, and third motion compensation errors, and the first, second, and third candidate motion vectors are determined according to the comparison result. And selecting means for selecting one of them as a motion vector corresponding to the block to be processed.

The data storage medium according to the present invention (claim 12) can be stored in a computer as a program.
It stores a program for performing a process of detecting a motion vector corresponding to a block to be processed by the described motion vector detection method.

The data storage medium according to the present invention (claim 13) can be stored in a computer as a program.
It stores a program for performing a process of detecting a motion vector corresponding to a block to be processed by the described motion vector detection method.

[0065]

Embodiments of the present invention will be described below. Embodiment 1 First, the basic concept of Embodiment 1 of the present invention will be described. Since an image signal generally has a strong correlation between local pixel values, a motion vector and a motion compensation error are almost equal between a target block and a neighboring block located adjacent to the target block on the processing target screen.
As a result, when the pixel value correlation is strong, the motion vector search center information for obtaining the motion vector for the target block is predicted and generated from the motion vector of the neighboring block, and the predicted motion vector search center information is generated as the target block. , The detection accuracy of the motion vector of the target block hardly deteriorates.

However, when the target block and the neighboring block each include a pixel of an object that moves differently, the correlation of the pixel values between the target block and the neighboring block is lost, and the motion vector and the motion compensation error are lost. Also differs greatly between the target block and the neighboring blocks.

Therefore, the motion vector search center information of the target block is predicted and generated from the motion vector of the neighboring block, and the motion compensation error corresponding to the motion vector search center is compared with the motion compensation error corresponding to the neighboring block. By switching the motion vector detection processing according to the comparison result, it is possible to reduce the calculation amount in the motion detection processing for the target block without significantly impairing the detection accuracy of the motion vector of the target block. Specifically, according to the comparison result between the motion compensation error for the motion vector search center and the motion compensation error for the neighboring block, when the difference between the two motion compensation errors is small, the motion vector search center information corresponding to the target block is obtained. Is used as a motion vector for the target block. On the other hand, when the difference between the two motion compensation errors is large, the motion vector obtained by performing the motion vector detection process with the motion vector search center of the target block as the search start position is used as the motion vector for the target block.

When the motion compensation error of the target block with respect to the motion vector search center is smaller than the motion compensation error of the neighboring block, the pixel value correlation between the neighboring block and the target block is low. Although the predicted motion vector search center may be different from the optimal motion vector of the target block, the motion compensation of the target block may be performed even if the position information indicating the motion vector search center is selected as the motion vector of the target block. Since the error is small, it can be considered that there is no practical problem.

Therefore, when the difference between the two motion compensation errors is smaller than a predetermined value, a determination result (discontinuation determination of motion detection) is output using the position information indicating the motion vector search center as the motion vector of the target block. Instead, when the motion compensation error with respect to the search center of the target block is smaller than the motion compensation error of the neighboring block, the position information indicating the motion vector search center is used as the motion vector of the target block (discontinuation determination of motion detection). A signal may be output.

FIG. 1 is a block diagram for explaining a motion vector detecting device according to the first embodiment of the present invention. The motion vector detecting device 101 according to the first embodiment detects a motion vector corresponding to a target block by a two-layer search process. In this device, a first-layer search process is input from outside. This is to calculate the motion compensation error for the motion vector search center information Sc.
For this reason, the motion estimation device 101 does not include the motion vector generator u1a and the minimum error selector u4a in the conventional hierarchical motion vector detection device 201 shown in FIG.

The motion vector detecting apparatus 101 according to the first embodiment includes a frame memory u2 in which image data of a reference screen is stored, and a specific block on the reference screen indicated by information Sc indicating a motion vector search center from outside. An error calculator u3a for calculating an error between the image data and the image data of the target block is provided. In the motion vector detection device 101, the frame memory u2 and the error calculator u3a constitute a first-layer motion detection unit.

The motion vector detecting device 101
Are read from the frame memory u2 based on the candidate motion vector MVc based on the MV generator u1 that generates a plurality of candidate motion vectors MVc using the pixel position indicated by the external motion vector search center information Sc as the search center. Motion compensation data (second motion compensation data) C2t
And an error calculator u3b that calculates error data D2p between the input target block image data St and a plurality of candidate motion vectors MVc based on the error data D2p.
Among them, a minimum error selector u4 for selecting a candidate motion vector corresponding to the minimum error as the second hierarchical motion vector MVs. This motion vector detecting device 10
In No. 1, the frame memory u2, the MV generator u1, the error calculator u3b, and the minimum error selector u4 constitute a second-layer motion detecting unit.

Further, the motion vector detecting device 101
Selects one of the error data D1p output from the error calculator u3a and the minimum error data Dmp output from the minimum error selector u4 based on the control signal Dc, and selects the selected data as the selected error data. A first selector u6 that outputs as Dsp, and motion vector search center information Sc input from the outside based on the control signal Dc.
And a second selector u7 that selects one of the second hierarchical motion vector MVs output from the minimum error selector u4 and outputs the selected motion vector as the motion vector MV of the target block. I have.

Here, the selector u6 has its first input node 6a receiving the minimum error data Dmp from the minimum error selector u4, and has its second input node 6b
Is the error data D1p output from the error calculator u3a.
Is input, and one of these data is selected error data D
The signal is output from the output node 6c as sp. The selector u7 has its first input node 7a receiving the second hierarchical motion vector MVs from the minimum error selector u4, and its second input node 7a.
The motion vector search center information Sc from the outside is input to b, and one of these data is output from the output node 7c as the motion vector MV of the target block.

The motion vector detecting device 101
Is a temporary storage memory u8 for temporarily storing the selection error data Dsp, a motion compensation error Drp of a neighboring block read from the temporary memory u8, and an error calculator u3a.
Is compared with the motion compensation error D1p corresponding to the motion vector search center calculated in step (1), and a discontinuation determination result indicating whether or not to terminate the motion vector search processing by the second-layer motion detector is output as the control signal Dc. And a discontinuation determiner u5a. Other configurations are the same as those of the conventional motion vector detection device 201.

FIG. 2A is a diagram showing a specific configuration of the discontinuation judging unit u5a constituting the motion vector detecting device 101 shown in FIG. The discontinuation determiner u5a calculates the motion compensation error Drp of the neighboring block and the motion compensation error corresponding to the motion vector search center Sc of the target block (FIG. 1).
In this example, the comparator c1a compares the motion compensation error with the output of the error calculator u3a. The comparator c1a outputs a determination signal Dc indicating a termination determination result according to the magnitude relationship between the two motion compensation errors. Note that in FIG. 2A, i10, i1
Numeral 1 is an input terminal of the decision unit u5a. The input terminal i10 receives the motion compensation error Drp of the neighboring block, and the input terminal i11 receives the motion compensation error D1p corresponding to the motion vector search center Sc. Reference numeral o10 denotes an output terminal of the determination unit u5a, and the output terminal o10 outputs the determination signal Dc.

That is, when the motion compensation error D1p with respect to the motion vector search center is smaller than the motion compensation error Drp with respect to the neighboring block, the discontinuation determiner u5a outputs the determination signal Dc indicating that the search processing of the second hierarchy is to be terminated. Then, the motion compensation error D1p for the motion vector search center becomes the motion compensation error D1 for the neighboring block.
When the speed is equal to or more than rp, a determination signal Dc indicating that the search processing of the second hierarchy is performed is output.

In the motion vector detecting apparatus 101 according to the first embodiment, the decision signal Dc indicating that the search processing of the second hierarchy is to be terminated is generated by the discontinuation decision unit u5a.
, The arithmetic processing in the error calculator u3b, the minimum error selector u4, and the MV generator u1 is stopped, and the selector u6 selects the error data D1p from the error calculator u3a. In the selector u7, the motion vector search center information Sc input from the outside is selected. On the other hand, when the decision signal Dc indicating that the search processing of the second hierarchy is to be performed is output from the discontinuation determiner u5a, the error calculator u3b,
The arithmetic processing is performed by the minimum error selector u4 and the MV generator u1, and the selector u6 selects the minimum error data Dmp from the minimum error calculator u4, and the selector u7.
, The second-layer motion vector MVs from the minimum error calculator u4 is selected.

Next, the function and effect will be described. Below
The operation of the first motion vector detection device 101 will be described. The temporary memory u8 stores a motion compensation error (error data) Drp corresponding to a neighboring block processed immediately before the target block to be processed, and the discontinuation determination unit u5a reads out the temporary memory u8. The motion compensation error Drp of the neighboring block and the motion compensation error D1p calculated by the error calculator u3a and corresponding to the motion vector search center information Sc of the target block are compared.
An abort determination result indicating whether or not to terminate the second-layer motion vector search process is output as a determination signal Dc.

More specifically, in the censoring judging device u5a, the motion compensation error D1p
Is smaller than the motion compensation error Drp for the neighboring block, a determination signal Dc indicating that the search processing of the second hierarchy is to be terminated is output. On the other hand, when the motion compensation error D1p for the motion vector search center is equal to or greater than the motion compensation error Drp for the neighboring block, a determination signal Dc indicating that the search processing of the second hierarchy is not terminated is output.

When it is determined by the above-mentioned discontinuation judging device u5a that the search processing of the second layer is to be discontinued, the motion vector generator u1, the error calculator u3b and the minimum error selector u5
4 is notified by the determination signal Dc that the arithmetic processing is unnecessary, and the error calculator u3b, the minimum error selector u4, and the MV generator u1 perform the arithmetic processing for the second hierarchical search processing. I can't. On the other hand, when it is determined that the search processing of the second hierarchy is not to be aborted by the abort determination unit u5a, the determination signal Dc is sent to the motion vector generator u1, the error calculator u3b, and the minimum error selector u4.
Informs that the arithmetic processing is necessary, the error calculator u3b, the minimum error selector u4, and the MV generator u
In 1, the arithmetic processing for the search processing of the second hierarchy is performed.

At this time, when the decision signal Dc indicates termination of the search processing of the second hierarchy, the selector u7 selects the motion vector search center information Sc and outputs it as the motion vector MV of the target block. . On the other hand, when the determination signal Dc indicates that the search processing of the second hierarchy is not terminated, the motion vector MVs selected by the minimum error selector u4 is output as the motion vector MV of the target block.

Similarly, in the selector u6, the judgment signal D
When c indicates the termination of the search processing of the second hierarchy, the motion compensation error D1 corresponding to the motion vector search center information Sc
p is selected and output to the temporary memory u8. On the other hand, if the determination signal Dc indicates that the search processing of the second hierarchy is not terminated, the motion compensation error Dmp corresponding to the motion vector MVs selected by the minimum error selector u4 is output to the temporary memory u8. Is done.

In the first embodiment having such a configuration, the censoring judging unit u5a refers to the motion compensation error Drp corresponding to the detected motion vector of the neighboring block to obtain the second
Since it is determined that the coding efficiency is not improved even if the hierarchical motion search is performed, unnecessary second-layer motion search processing can be stopped, thereby reducing the number of error calculations in the motion vector detection processing. Can be.

Further, when the motion compensation error D1p of the motion vector search center of the target block is smaller than the motion compensation error Drp of the neighboring block, the censoring judging device u5a converts the motion vector search center information Sc into the motion vector of the target block. In this case, the pixel value correlation between the neighboring block and the target block is low, and the motion vector search center predicted from the neighboring block may be different from the optimal motion vector of the target block. Motion compensation error D for vector search center
Since 1p is small, it can be considered that there is no practical problem.

In the first embodiment, the determination as to whether or not to terminate the search processing of the second hierarchy is made based on the motion compensation error D1p with respect to the motion vector search center of the target block.
This is performed in accordance with the magnitude relationship between the motion compensation error Drp and the motion compensation error Drp for the neighboring block. However, if the magnitude of the difference between the motion compensation error D1p and the motion compensation error Drp is less than a predetermined value, the second
The search process of the second hierarchy may be terminated, and if the difference is equal to or more than a predetermined value, the search process of the second hierarchy may not be terminated.

FIG. 2B shows the structure of the discontinuation judging device u51a having such a structure. The termination determination unit u51a includes a reference value generator c1b for generating the predetermined reference value Dcp, a motion compensation error D1p, and a motion compensation error D1.
a comparison / decision unit c51a that compares the magnitude of the difference of rp with the predetermined reference value Dcp and outputs a decision signal Dc. When the magnitude of the difference is smaller than a predetermined value, the comparison / determination unit c51a outputs a determination signal Dc indicating that the search processing of the second hierarchy is terminated. Is configured to output a determination signal Dc indicating that the search processing of the second hierarchy is not terminated.

(Embodiment 2) FIG. 3 is a block diagram for explaining a motion vector detecting apparatus according to Embodiment 2 of the present invention. The motion vector detecting apparatus 102 according to the second embodiment includes a motion compensation error Drp corresponding to a detected motion vector of a neighboring block and a motion compensation error Drp of the target block, instead of the censoring determiner u5a in the motion vector detecting apparatus according to the first embodiment. Censoring determination that outputs a determination signal Dcb indicating whether or not to terminate the second-layer motion search based on the magnitude relationship with the motion compensation error D1p corresponding to the motion vector search center and the threshold value Th input from the outside. U
5b. Other configurations of the motion vector detecting device 102 are the same as those of the motion vector detecting device 101 of the first embodiment. In FIG. 3, i3
Is a motion vector detecting device 1 to which the threshold value Th is input.
02 is a third input terminal.

FIG. 4 is a diagram showing a specific configuration of the discontinuation judging unit u5b constituting the motion vector detecting device 102 shown in FIG. The discontinuation determination unit u5b compares the motion compensation error Drp corresponding to the neighboring block read from the temporary memory u8 with the motion compensation error D1p corresponding to the motion vector search center information Sc calculated by the error calculator u3a. 1 comparator c2a, and a second comparator c2b for comparing the threshold value Th input from the outside with the motion compensation error D1p corresponding to the motion vector search center information Sc.
It is composed of

Here, the threshold value Th corresponds to the target block, which is determined by the bit rate of the encoded data output from the inter-frame motion compensation encoding device and the number of motion compensation error calculations performed by the motion vector detecting device. Is the allowable value of the motion compensation error. Further, the first comparator c2a
Is the motion compensation error D1p with respect to the motion vector search center.
Is smaller than the motion compensation error Drp for the neighboring block, a determination signal Dca indicating that the search processing of the second layer is to be terminated is output. If the motion compensation error D1p for the motion vector search center is greater than or equal to the motion compensation error Drp for the neighboring block, At one time, the configuration is such that a determination signal Dca indicating that the search processing of the second hierarchy is performed is output. The second comparator c2b determines that the motion compensation error D1p with respect to the motion vector search center is equal to the threshold Th.
If it is less than the above, the determination signal Dca output from the first comparator c2a is output as it is as the determination signal Dcb, and the motion compensation error D1p with respect to the motion vector search center is obtained.
Is greater than or equal to the threshold Th, the first comparator c2a
Irrespective of the determination signal Dca from, the determination signal Dcb indicating that the search processing of the second hierarchy is not terminated (that is, the search processing of the second hierarchy is performed) is output. In FIG. 4, i12 is a third input terminal of the discontinuation determiner u5b to which the threshold value Th from the outside of the motion vector detecting device 102 is input.

Next, the function and effect will be described. First, the basic principle of the second embodiment will be briefly described. In the first embodiment, the motion compensation error Drp corresponding to the detected motion vector of the neighboring block and the motion compensation error D1 corresponding to the motion vector search center of the target block are simply calculated.
p is compared to generate a determination signal Dc indicating a censoring determination result. However, when the accuracy of motion detection for a nearby block is low and a motion compensation error corresponding to a nearby block is large, Even if the motion compensation error obtained by performing motion compensation on the target block based on the motion vector search center information Sc is large, the motion detection processing of the second layer is terminated, and the motion vector search center with poor accuracy is obtained. The information Sc is output as the motion vector MV of the target block.

Therefore, for example, the allowable value of the motion compensation error determined by the bit rate of the encoded data output from the inter-frame motion compensation encoding device and the number of motion compensation error calculations in the motion vector detecting device is set to a threshold value. The motion compensation for the target block is provided to the above-described censoring determiner u5b as Th, so that the motion vector search center information Sc
Is obtained based on the threshold Th.
If it is larger, the termination of the motion detection processing of the second layer can be stopped.

Next, the operation will be described. In the motion vector detection device 102 according to the second embodiment, the operation other than the termination determination unit u5b is the same as the operation of the motion vector detection device 101 according to the first embodiment. I will explain only.

In the termination determination unit u5b, the temporary memory u
A comparison is made between the motion compensation error Drp corresponding to the neighboring block read from No. 8 and the motion compensation error D1p corresponding to the motion vector search center calculated by the error calculator u3a, and further input from the outside. The threshold Th is compared with the motion compensation error D1p corresponding to the motion vector search center information Sc.

Then, the discontinuation judging device u5b determines the second
A termination determination signal Dcb indicating whether or not to terminate the motion detection of the hierarchy is output. If you decide to discontinue,
Based on the determination signal Dcb, the motion vector generator u1,
The error calculator u3b and the minimum error selector u4 are notified that no processing is required.

Hereinafter, the specific operation of the above-described discontinuation judging device u5b will be described. First, the above-mentioned termination judgment unit u
5b, the comparator c2a compares the motion compensation error Drp corresponding to the neighboring block and the motion compensation error D1p corresponding to the motion vector search center of the target block, as in the discontinuation determiner u5a of the first embodiment. You. Then, the motion compensation error D1p for the motion vector search center
Is smaller than the motion compensation error Drp for the neighboring block, the comparator c2a outputs a determination signal Dca indicating that the search processing of the second hierarchy is to be terminated. On the other hand, the motion compensation error D1p for the motion vector search center
Is greater than or equal to the motion compensation error Drp for the neighboring block, the comparator c2a outputs a determination signal Dca indicating that the search processing of the second hierarchy is not terminated.

At this time, the comparator c2b compares the motion compensation error D1p corresponding to the center of the motion vector search of the target block with the above threshold value Th, and this motion compensation error D1
If 1p is less than the threshold Th, the output Dc of the comparator c2a
a is output as it is as the termination determination signal Dcb,
If the motion compensation error D1p is equal to or greater than the threshold Th, the second-layer motion search process is performed (ie, the second-layer search process is not terminated) regardless of the determination output Dca of the comparator c2a. Is output.

In the second embodiment having such a configuration, the motion compensation error D1p based on the threshold value Th input from the outside and the motion vector search center information Sc in the discontinuation determiner u5b.
When the motion compensation error obtained by performing motion compensation on the target block based on the motion vector search center information Sc is larger than the threshold, the motion search processing of the second hierarchy is always performed. , The accuracy of motion detection can be improved.

In the second embodiment, when the motion compensation error D1p of the target block with respect to the motion vector search center is smaller than the motion compensation error Drp of the neighboring block, the comparator c2a of the above-mentioned discontinuation judging unit u5b is used. Is output, and the motion compensation error D1p of the target block with respect to the motion vector search center is changed to the motion compensation error Dr of the neighboring block.
When the value is equal to or larger than p, the comparator c2a outputs a determination signal Dca indicating that the detection processing of the second hierarchy is to be performed. When the difference between the error D1p and the motion compensation error Drp is smaller than a predetermined reference value Dcp, a determination signal Dc indicating that the search processing of the second hierarchy is to be terminated.
a, and when the difference is equal to or greater than the predetermined value Dcp, a determination signal Dca indicating that the search processing of the second hierarchy is not terminated may be output.

(Embodiment 3) FIG. 5 is a block diagram for explaining a motion vector detecting device according to Embodiment 3 of the present invention. The motion vector detecting apparatus 103 according to the third embodiment is different from the motion vector detecting apparatus according to the second embodiment in that the censoring determiner u5b is used instead of the motion compensation error Drp corresponding to the detected motion vector of the neighboring block. The magnitude relationship with the motion compensation error D1p corresponding to the motion vector search center, the threshold value T input from the outside
h, and a determination signal D indicating whether or not to terminate the second-layer motion search based on the second-layer motion detection command signal Cm.
It has a censoring judging device u5c that outputs cf, and the other configuration is the same as that of the motion vector detection device 102 of the second embodiment. In FIG. 5, i4 is a fourth input terminal of the motion vector detection device 103 to which the second-layer motion detection command signal Cm is input.

FIG. 6 is a diagram showing a specific configuration of the termination determination unit u5c constituting the motion vector detecting device 103 shown in FIG. This discontinuation determiner u5c is a comparator c2 in the discontinuation determiner u5b of the second embodiment.
In addition to comparators c3a and c3b having the same configuration as a and c2b, a signal generator c3c for generating an abort inhibition signal Fc
And a second-layer motion detection command signal Cm input from the outside
Selector c3 for selecting one of the output Dcb of the second comparator c3b and the termination prohibition signal Fc based on
And

Here, when the second-layer motion detection command signal Cm is input from outside, the selector c3 selects the termination prohibition signal Fc, and receives the second-layer motion detection command signal Cm. When not performed, the output Dcb of the second comparator c3b is selected. The device external to the motion vector detection device 103 that outputs the second-layer motion detection command signal Cm includes, for example, an intra-screen / inter-screen encoding mode determination for determining a coding mode for a target block or a neighboring block. Vessels and the like.

Next, the function and effect will be described. First, the basic principle of the third embodiment will be briefly described. When the neighboring block located in the vicinity of the target block is a block having no motion vector, that is, a block to be processed first or a block at the end of the screen, or the neighboring block is subjected to intra-frame encoding processing. In the case of the intra block, the motion vector search center information Sc predicted and generated from the motion vector of the block located near the target block is considered to have low reliability. If the neighboring block is an intra block that has been subjected to the intra-frame encoding process, the motion compensation error for the target block is large, and the motion vector of the target block is often unreliable.

Therefore, in such a situation, the second-layer motion detection command signal C from a device such as an intra / inter-frame coding mode determiner outside the motion vector detecting device 103 is used.
m instructs the discontinuation determiner u5c that the reliability of the motion vector search center Sc is low, and outputs the determination signal Dcf indicating that the second-layer motion detection is to be performed, by the discontinuation determiner u.
By outputting 5c, the accuracy of motion detection can be further improved.

Next, the operation will be described. In the motion vector detecting apparatus 103 according to the third embodiment, all operations other than the above-described discontinuation determiner u5c are performed in the second embodiment shown in FIG.
Since the motion vector detecting device 102 is the same as the motion vector detecting device 102, only the operation of the censoring determiner u5c will be described below.
In the termination determination unit u5c shown in FIG.
Similarly to the truncation determination unit u5b, the comparator c3a compares the motion compensation error Drp corresponding to the neighboring block with the motion compensation error D1p corresponding to the motion vector search center of the target block. The comparator c3b compares the threshold value Th obtained with the motion compensation error D1p corresponding to the motion vector search center.

Then, the censoring judging device u5c determines the motion vector of the second layer based on the comparison results of the comparators c3a and c3b and whether or not the second-layer motion detection command signal Cm is input. A termination determination signal Dcf indicating whether to terminate the search is output. When it is determined that the processing is to be discontinued, the determination signal Dcf notifies the motion vector generator u1, the error calculator u3b, and the minimum error selector u4 that no processing is required.

Hereinafter, the specific operation of the above-mentioned termination judgment unit u5c will be described. First, the above-mentioned termination judgment unit u
In 5c, the motion compensation error Drp corresponding to the neighboring block and the motion compensation error D1p corresponding to the motion vector search center of the target block are compared by the comparator c3a, similarly to the discontinuation determination unit u5a of the second embodiment. You. Then, the motion compensation error D1p for the motion vector search center
Is smaller than the motion compensation error Drp for the neighboring block, the comparator c3a outputs a determination signal Dca indicating that the search processing of the second hierarchy is to be terminated.
On the other hand, the motion compensation error D1 with respect to the motion vector search center
When p is equal to or greater than the motion compensation error Drp for the neighboring block, the comparator c3a outputs a determination signal Dca indicating that the search processing of the second hierarchy is not terminated.

At this time, the comparator c3b compares the motion compensation error D1p corresponding to the motion vector search center of the target block with the threshold Th. As a result of this comparison, if the motion compensation error D1p corresponding to the motion vector search center is less than the threshold Th, the output Dca of the comparator c3a is output as it is as the discontinuation determination signal Dcb. on the other hand,
Motion compensation error D1 corresponding to the above motion vector search center
If p is equal to or greater than the threshold Th, the second-layer motion vector search is performed regardless of the determination output Dca of the comparator c3a (that is, the second-layer search process is not terminated).
Is output.

When the second-layer motion detection command signal Cm is input to the selector c3, the selector c3
Then, the termination prohibition signal Fc is selected and output as the final determination data Dcf. Conversely, when the second-layer motion detection command signal Cm is not input to the selector c3, the selector c3 selects the determination signal Dcb from the comparator c3b and outputs it as the final determination data Dcf. .

As described above, in the third embodiment, the second-layer motion detection command signal Cm from the outside forcibly causes the second
Since the decision result indicating that the hierarchical motion vector search is to be performed is output from the discontinuation determiner u5c, it is possible to further improve the motion vector detection accuracy as compared with the motion vector detection device 102 according to the second embodiment shown in FIG. it can.

In the third embodiment, when the motion compensation error D1p corresponding to the motion vector search center of the current block is smaller than the motion compensation error Drp of the neighboring block,
A decision signal Dca indicating that the motion vector search processing of the hierarchical layer is to be terminated is output. When the motion compensation error D1p corresponding to the motion vector search center of the target block is equal to or greater than the motion compensation error Drp of the neighboring block, the second hierarchical layer The comparator c3a is configured to output a determination signal Dca indicating that the motion vector search processing is performed.
Similarly to the comparator c51a shown in (b), the motion compensation error D1p
When the magnitude of the difference between the motion compensation error Drp and the motion compensation error Drp is smaller than the predetermined reference value Dcp, the determination signal Dca indicating that the motion vector search processing of the second hierarchy is to be terminated is output, and the difference is equal to or greater than the predetermined value Dcp. In the case of, the configuration may be such that the determination signal Dca indicating that the motion vector search processing of the second hierarchy is not terminated is output.

Also, in the first to third embodiments, a specific example of the second-layer motion vector search processing, that is, the processing of performing a motion vector search with the center of the motion vector search as a search start position and detecting a motion vector for the target block. Although the detailed contents are not shown, in the second-layer motion vector search processing, the conventional three-step method shown in FIG. 13, another conventional hierarchical motion vector detection method shown in FIG. 14, or the Oneat method shown in FIG. A motion vector detection method such as a time is used.

(Embodiment 4) Next, a motion vector detecting apparatus according to Embodiment 4 of the present invention will be described. First, the basic principle of the motion vector detecting device according to the fourth embodiment will be described. FIG. 7 is a diagram for explaining the principle of operation of the motion vector detection device according to the present embodiment.
The pixel on the reference screen referred to at the time of is shown.

In FIG. 7, P0 is a reference pixel at a position indicated by the motion vector found by the search processing of the first hierarchy, and surrounding pixels P1 to P8 are located around this reference pixel P0. . These peripheral pixels P1 to P
The position of 8 is a virtual pixel position indicated by a motion vector having an accuracy higher than an integer pixel accuracy such as 画素 pixel accuracy, and the pixel values of these pixels are the pixel values of actually existing pixels. It is generated by interpolation.

FIG. 8 is a diagram for explaining the operation of the motion vector detecting device according to the fourth embodiment, and shows an example of a specific method for performing interpolation generation of pixel values. In FIG. 8, Q1, Q2, Q3, and Q4 are actual pixels constituting a reference screen, and are located at positions corresponding to vertices of a predetermined square on the reference screen. The pixels Q1, Q2, Q3, and Q4 have pixel values q1 and q, respectively.
2, q3 and q4.

H1 is a virtual pixel located between the pixels Q1 and Q2, H2 is a virtual pixel located between the pixels Q1 and Q3, and H3 is a virtual pixel located between the pixels Q1 and Q4. is there. These virtual pixels H1, H
Generally, the pixel values h1, h2, and h3 of H2 and H3 are generally calculated by the following equations (1), (2), and (3), respectively. h1 = (q1 + q2) / 2 (1) h2 = (q1 + q3) / 2 (2) h3 = (q1 + q2 + q3 + q4) / 4 (3)

As described above, in the motion detection processing with higher precision than the integer pixel precision (high pixel precision motion detection), the pixel value of a virtual pixel virtually placed at a position where no pixel exists on the reference screen is determined. Must be generated by interpolation processing, so that the amount of calculation increases as compared with the motion detection processing with integer pixel accuracy. However, on the other hand, since the motion vector obtained by the high-pixel-accuracy motion detection process can accurately represent even a small motion of a small image, the motion compensation error using such a motion vector causes a motion compensation error. Become smaller. Further, since the above-described pixel value interpolation processing is equivalent to a low-pass filter, it is also known that the interpolation processing has an effect of suppressing high frequency noise.

Since the high frequency noise suppression effect cannot be ignored in the motion detection processing with higher accuracy than the integer pixel accuracy, for example, the motion compensation error caused by the motion vector indicating the pixel P0 in FIG. , P7
Pixel P1, P3,
The motion compensation error due to the motion vectors indicating P6 and P8 is often smaller than the motion compensation error corresponding to the pixel P0.

However, the pixels P1, P3, P6, P
If the motion detection is performed for all of 8, the number of error calculations increases. Therefore, in the search processing of the second hierarchy, the pixel P2,
The motion compensation error is calculated for the four motion vectors of the reference pixels indicating the positions of P4, P5, and P7, and the pixels P2, P
4, P5, and P7, the pixel position Pi at which the motion compensation error is minimized is detected.
The search is limited to two motion vectors indicating the pixel positions adjacent to the pixel position Pi among 1, P3, P6, and P8.

Finally, the first, second, and third layers
The motion compensation errors corresponding to the motion vectors obtained by the hierarchical search process are compared, and the motion vector with the smallest motion compensation error is set as the motion vector of the target block. Although this method is similar to the conventional hierarchical three-step method, the conventional three-step method selects one motion vector that minimizes the motion compensation error for each layer. The method of (1) selects one motion vector with the minimum motion compensation error from the motion vectors obtained by the search processing of each layer.

For example, in the conventional three-step method, the motion compensation error corresponding to the position of the pixel P0 shown in FIG. 7 is larger than the motion compensation error corresponding to the positions of the pixels P2, P4, P5, and P7 shown in FIG. If it is smaller, a motion vector indicating the position of the pixel P0 will be selected in the second hierarchy, and in the search processing in the third hierarchy, it is determined that the search range is narrowed down from the four candidates of the pixels P1, P3, P6, and P8. There will be no criteria. On the other hand, in the present embodiment, since the pixel P0 indicated by the motion vector obtained by the search of the first hierarchy is not considered in the search of the second hierarchy, the pixels P2, P2
One of 4, P5, and P7 is detected as the pixel position Pi where the motion compensation error is minimized. Therefore,
In the search processing of the third hierarchy, the search range can be narrowed from the four candidates of the pixels P1, P3, P6, and P8 based on the pixel position Pi as a criterion.

FIG. 9 is a block diagram for explaining a motion vector detecting device according to the fourth embodiment of the present invention. The motion vector detecting device 104 according to the fourth embodiment includes a motion vector MV of the second hierarchy selected by the minimum error selector u4b in addition to the configuration of the conventional motion vector detecting device 201.
2s, a motion vector generator u1c that generates a plurality of candidate motion vectors MV3c according to the motion vector MV1s of the first hierarchy selected by the minimum error selector u4a, and the candidate vector read from the frame memory u2. An error calculator u3 that calculates an error between the image data C3t corresponding to the specific block of the reference screen specified by the motion vector MV3c and the image data Sc of the input target block.
c, from among the three motion vectors corresponding to each layer obtained by the search processing of the first, second, and third layers, a motion vector that minimizes the motion compensation error is selected;
A minimum error selector u4c that outputs the selected motion vector as a motion vector MV3s corresponding to the target block
And The motion vector detection device 10
In No. 4, the MV generator u1c, the error calculator u3c, the minimum error selector u4c, and the frame memory u2 constitute a third-layer motion vector detecting unit.

Here, the motion vector generator u1c
Is a pixel position in a third search range that is not limited by the first-layer search range and the second-layer search range and is limited by the first-layer motion vector MV1s and the second-layer motion vector MV2s. The configuration is such that the candidate motion vector MV3c shown is detected. Further, the minimum error selector u
4c includes the first and second signals from the minimum error selectors u4a and u4b.
The second selected motion vector MVs1, MVs2 and the first,
The second motion compensation errors D1pm and D2pm and a plurality of candidate motion vectors MV3c from the motion vector generator u1c
And the motion compensation error D3p from the error calculator u3c. In the minimum error selector u4c, 3
A process of selecting a motion vector by searching for a layer;
The process of selecting a motion vector with the smallest motion compensation error from among the three motion vectors corresponding to each layer selected by the search process of the second, third, and third layers is collectively performed. It has become.

In the fourth embodiment, the motion vector detecting section of the second hierarchy comprising the MV generator u1b, the error calculator u3b, the minimum error selector u4b, and the frame memory u2 performs the search of the first hierarchy. The configuration is such that the vicinity of the pixel position indicated by the motion vector MV1s obtained by the processing is set as the second search range, and the motion vector MV2s of the second hierarchy is detected. That is, the minimum error selector u4b
Is the minimum error selector u of the motion vector detecting section of the first layer.
Without considering the motion vector MV1s output from the MV generator 4a, the motion vector with the minimum motion compensation error is selected from the candidate motion vectors MV2c generated by the MV generator u1b. MV2s is selected. Here, the pixel position indicated by the motion vector MV1s is the position of the pixel P0 in FIG. 7, and the second search range is the pixel P2,
This is an image space including P4, P5, and P7. Then, if the motion vector MV2s of the second hierarchy indicates the position of the pixel P7, the third search range is the pixel P6 in FIG.
, And P8.

Next, the operation will be described. In the motion vector detecting apparatus 104 according to the fourth embodiment, the first hierarchical search processing is the same as the processing in the conventional hierarchical motion vector detecting apparatus 201.
The search processing of the first, second, and third layers will be described.

In the motion vector generator u1a in the motion vector detecting device 104, a motion vector indicating a plurality of pixel positions near the search center is determined as a candidate motion vector based on vector search center information Sc supplied from outside the device. Output as MV1c. Then, the frame memory u2 outputs an image signal (motion compensation data) C1t corresponding to the specific block on the reference screen indicated by the candidate motion vector MV1c. In the error calculator u3a, a difference signal (image error data) D1 between the motion compensation data C1t and the image signal St1 of the target block.
p is calculated. In the minimum error selector u4a, the image error data D1p is monitored, and a candidate motion vector MV1c with the minimum error data D1p is selected from a plurality of candidate motion vectors MV1c, and the selected motion vector (first hierarchical level) is selected. Is output as MV1s. For example, the selected motion vector (first-layer motion vector) MV1s indicates the position of the pixel P0 in FIG.

When the first-layer motion vector MV1s is input to the second-layer motion detector of the motion vector detecting device 104, the MV generator u1b of the second-layer motion detector detects the first-layer motion vector MV1s. Are generated, a plurality of second candidate motion vectors MV2c corresponding to positions near the pixel positions on the reference screen indicated by. These second candidate motion vectors MV2c are, for example, pixels P2 and P2 in FIG.
4, P5 and P7 are shown.

At this time, the image data (motion compensation data) C2t corresponding to the specific block on the reference screen indicated by the second candidate motion vector MV2c read from the frame memory u2 is input to the error calculator u3b. , The motion compensation data C2t and the image data St of the target block.
Is calculated as error data D2p.

The minimum error selector u4b has a plurality of candidate motion vectors MV generated by the MV generator u1b.
Error data D2p corresponding to each candidate motion vector MV2c is input together with the candidate motion vector MV2c, and a candidate motion vector MV2c with the minimum error data is selected and output as a second-layer motion vector (selected motion vector) MV2s. For example, the motion vector (selected motion vector) MV2s of the second hierarchy indicates the position of the pixel P7 in FIG.

Further, the motion vector generator u1c of the third-layer motion detecting section supplies the motion vector MV2s of the second hierarchy from the minimum error selector u4b and the minimum error selector u4a.
When the motion vector MV1s of the first hierarchy from is input, the motion vector generator u1c restricts the motion vectors MV1s and MV2s to the first
And a plurality of candidate motion vectors MV3c corresponding to pixels in the third search range not including the second search range. For example, the plurality of candidate motion vectors MV3c
Shows the positions of the pixels P6 and P8 in FIG.

Then, a plurality of candidate motion vectors MV3c
Is stored in the frame memory u2.
And the error calculator u3c calculates the motion compensation data C3t corresponding to the candidate motion vector MV3c,
Error data D3p between the input target block and the image data St is calculated.

Further, in the minimum error selector u4c, the selected motion vectors MV1s and MV2s corresponding to the target block obtained by the search processing of the first and second layers,
In addition, among the candidate motion vectors MV3c obtained by the search processing of the third hierarchy, the motion vector with the minimum motion compensation error is selected and output as the motion vector MV3s of the target block.

As described above, in the fourth embodiment, the motion vector MV2s, which is the motion vector indicating the pixel position outside the range searched in the first hierarchy in the second hierarchy and has the minimum motion compensation error, is detected. The search range of the third layer is determined based on the two motion vectors detected by the search of the first and second layers, the motion vector obtained by the search processing of the third layer, and the first and second layers , The motion vector corresponding to the target block is detected from among the motion vectors obtained by the search of, so that the search range of the third layer can be efficiently narrowed down, and the number of error calculations in the motion vector detection process can be reduced. Reduction can be realized.

In the fourth embodiment, in the second search range, the position of the prediction block is set with the pixel position indicated by the first selected motion vector MV1s as the search center.
The search is performed along the vertical direction and the horizontal direction on the reference screen, and in the third search range, the position of the prediction block is set in the horizontal direction on the reference screen with the pixel position indicated by the second selected motion vector MV2s as the search center. Although the search is performed only along, the search directions in the second and third search ranges are not limited to this. For example, in the second search range, the position of the prediction block is searched only along the vertical or horizontal direction on the reference screen, with the pixel position indicated by the first selected motion vector MV1s as the search center. In the search range of, the position of the prediction block may be searched along a direction other than the search direction in the second search range, with the pixel position indicated by the second selected motion vector MV2s as the search center.

Further, by recording a motion detection program for realizing the motion vector detection processing by the motion vector detection device shown in each of the above embodiments on a storage medium such as a floppy disk, The motion vector detection processing described in the embodiment can be easily performed in a computer system.

FIG. 10 illustrates a case where the motion vector detecting process by the motion vector detecting device according to the first to fourth embodiments is performed by a computer system using a floppy disk storing the motion detecting program. FIG. FIG. 10A shows the appearance of the floppy disk as viewed from the front, its sectional structure, and the floppy disk main body, and FIG. 10B shows an example of the physical format of the floppy disk main body. The floppy disk FD has a structure in which the floppy disk main body D is housed in a floppy disk case FC, and a plurality of tracks are formed on the surface of the floppy disk main body D concentrically from the outer periphery toward the inner periphery. Tr is formed, and each track Tr is divided into 16 sectors Se in the angular direction. Accordingly, in the floppy disk FD storing the program, the floppy disk main body D has the data as the program recorded in an area (sector) Se allocated thereon. FIG. 10C shows a configuration for recording the program on the floppy disk FD and performing image processing by software using the program stored in the floppy disk FD.

The above program is stored in a floppy (registered trademark).
When recording the data on the disk FD, the data as the above program is transmitted from the computer system Cs to the floppy disk FD via the floppy disk drive FD.
Write to. When the motion vector detecting device is constructed in the computer system Cs by the program recorded on the floppy disk FD, the program is read from the floppy disk FD by the floppy disk drive FDD and loaded into the computer system Cs.

In the above description, the description has been made using a floppy disk as a data recording medium. However, even when an optical disk is used, a motion vector detecting process using software can be performed as in the case of the floppy disk. . The data recording medium is not limited to this,
The present invention can be similarly implemented as long as the program can be recorded, such as a card or a ROM cassette.

[0139]

As described above, the present invention (claims 1, 2 and 2)
According to 3, 4, 5, 10, 12), a motion compensation error corresponding to a neighboring block located near a block to be processed is compared with a motion compensation error corresponding to a motion vector search center, and the two motion compensation errors are compared. According to the magnitude of the difference between these motion compensations obtained by comparing the compensation errors, the position information indicating the motion vector search center is output as the motion vector of the block to be processed, or the search processing and the motion of the prediction block are performed. A motion vector obtained by performing the vector detection process with the above-described motion vector search center as a search start position is output or switched as a motion vector of the block to be processed, so that the motion detection accuracy is hardly impaired. It is possible to reduce the number of error calculations with the largest amount of computation, thereby reducing power consumption. It can be realized le detection process.

The present invention (claims 6, 7, 8, 9, 1)
According to (1, 13), the search processing of the prediction block is performed in the first and second search ranges, and the first and second selected motion vectors in which the motion compensation error is minimized in each search range are detected. ,
A third search range is narrowed down based on the first and second selected motion vectors, and a third selected motion vector in which the motion compensation error is minimized in the third search range is detected.
The motion compensation errors corresponding to the second and third selected motion vectors are compared, and one of the first, second, and third candidate vectors is selected as a motion vector corresponding to the block to be processed. Therefore, the third search range can be efficiently narrowed, and the number of error calculations can be reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram for explaining a motion vector detection device according to a first embodiment of the present invention.

FIG. 2 is a diagram showing a configuration (FIG. (A)) of a discontinuation determiner and a configuration (FIG. (B)) of a modified example thereof, which constitute the motion vector detecting device of the first embodiment.

FIG. 3 is a block diagram for explaining a motion vector detection device according to a second embodiment of the present invention.

FIG. 4 is a diagram illustrating a configuration of a discontinuation determiner included in the motion vector detection device according to the second embodiment.

FIG. 5 is a block diagram for explaining a motion vector detection device according to a third embodiment of the present invention.

FIG. 6 is a diagram illustrating a configuration of a discontinuation determiner included in the motion vector detection device according to the third embodiment.

FIG. 7 is a diagram for explaining a basic principle of a motion vector detection device according to a fourth embodiment of the present invention.

FIG. 8 is a diagram for explaining a basic principle of a motion vector detection device according to the fourth embodiment.

FIG. 9 is a block diagram for explaining a motion vector detection device according to the fourth embodiment.

FIG. 10 illustrates a data storage medium (FIGS. (A) and (b)) storing a program for performing the motion detection processing of each of the embodiments by a computer system, and the computer system (FIG. (C)). FIG.

FIG. 11 is a block diagram for explaining a conventional typical inter-picture motion compensation coding apparatus.

FIG. 12 is a block diagram for explaining a typical motion detection device constituting a conventional inter-picture motion compensation coding device.

FIG. 13 shows a conventional motion vector detection method (three-step method).
FIG.

FIG. 14 is a diagram for explaining a conventional motion vector detection method (１ ／ pixel accuracy search).

FIG. 15 is a block diagram illustrating a conventional hierarchical motion vector detection device.

FIG. 16 shows a conventional motion vector detection method (One at a tim
It is a figure for explaining e).

[Explanation of symbols]

101-104 Motion vector detecting device u1a, u1b, u1c Motion vector generator u2 Frame memory u3a, u3b, u3c Error calculator u4a, u4b, u4c Minimum error selector u5a, u5b, u5c Discontinuation judging device u6, u7 selector u8 Temporary memory C1t First motion compensation data C2t Second motion compensation data Cm Motion detection command signal Dc Decision data Dca First decision data Dcb Second decision data Dcf Final decision data D1p, D2p, D3p First, second , Third image error data Dmp minimum error data Dsp selection error data Drp reference error data D1pm, D2pm first and second motion compensation errors (minimum error data) MVc candidate motion vector MVs selected motion vector MV1c, MV2c, MV3c 1st, 2nd, 2nd MVs Selected motion vector MV1s, MV2s First and second selected motion vectors (first and second hierarchical motion vectors) MV3s Motion vector of target block St Image signal Sc Motion vector search center information Et Image encoded data Cs Computer system D Floppy disk body FC Floppy disk case FD Floppy disk FD Floppy disk drive Se Sector Tr Track

Claims (13)

[Claims] 1. For each block of a predetermined number of pixels, which is a unit for processing image data corresponding to a screen to be processed,
A motion vector for searching for a position on a reference screen of a prediction block corresponding to a target block to be processed under a predetermined condition, and detecting information indicating the position of the prediction block as a motion vector corresponding to the target block. A motion vector detection method including a detection process, wherein image data between a block corresponding to a motion vector search center, which is a position where a search for a position of the prediction block is started, and a processing target block to be processed is performed. Calculating a motion compensation error corresponding to the motion vector search center, and a neighboring block located near the block to be processed,
A comparison process of comparing a motion compensation error corresponding to a neighboring block, which is an error of image data with a corresponding prediction block, with a motion compensation error corresponding to the motion vector search center. According to the comparison result of the motion compensation error, the position information indicating the motion vector search center is output as a motion vector corresponding to the block to be processed, or the motion vector detection processing is performed by searching the motion vector search center for the search start position. A motion vector detection method characterized by switching whether to output a motion vector obtained as a motion vector corresponding to a block to be processed. 2. The motion vector detecting method according to claim 1, wherein when the motion compensation error corresponding to the motion vector search center is smaller than the motion compensation error corresponding to the neighboring block, the position information indicating the motion vector search center. Is output as a motion vector corresponding to the block to be processed. 3. The motion vector detecting method according to claim 1, wherein the motion compensation error corresponding to the motion vector search center is smaller than the motion compensation error corresponding to the neighboring block, and the motion corresponding to the motion vector search center. A motion vector detecting method, wherein when the compensation error is less than a predetermined value, the position information indicating the motion vector search center is output as it is as a motion vector corresponding to the block to be processed. 4. The motion vector detection method according to claim 3, wherein when a motion detection command signal is input from outside, a motion compensation error corresponding to the neighboring block and a motion compensation error corresponding to the motion vector search center are calculated. Irrespective of the magnitude relation, outputting a motion vector obtained by performing the motion vector detection process using the motion vector search center as a search start position as a motion vector corresponding to the block to be processed. . 5. The motion vector detecting method according to claim 1, wherein a difference between the motion compensation error corresponding to the neighboring block and the motion compensation error corresponding to the motion vector search center is smaller than a predetermined value. And outputting the position information indicating the motion vector search center as a motion vector corresponding to the block to be processed. 6. A block comprising a predetermined number of pixels, which is a unit for processing image data corresponding to a screen to be processed,
A motion vector for searching for a position on a reference screen of a prediction block corresponding to a target block to be processed under a predetermined condition, and detecting information indicating the position of the prediction block as a motion vector corresponding to the target block. A detection method, wherein a position of the prediction block is searched in a first search range set on the reference screen, and a position on the reference screen specified by a first candidate motion vector corresponding to the block to be processed is detected. Calculating a first motion compensation error, which is an error in image data between the block of interest and the block to be processed on the screen to be processed, and a position near a pixel position indicated by the first candidate motion vector. In the second search range on the reference screen, the position of the prediction block is searched, and the reference specified by the second candidate motion vector corresponding to the block to be processed is searched. Calculating a second motion compensation error, which is an error of image data between the block on the screen and the block to be processed on the screen to be processed; and the first candidate motion vector and the second candidate motion In the third search range on the reference screen set based on the vector, the position of the prediction block is searched, and the position on the reference screen specified by the third candidate motion vector corresponding to the block to be processed is searched. Calculating the third motion compensation error, which is the error of the image data between the block of interest and the block to be processed on the screen to be processed, and comparing the first, second, and third motion compensation errors And selecting one of the first, second, and third candidate motion vectors as a motion vector corresponding to the block to be processed, according to the comparison result. Motion vector detection method. 7. The motion vector detection method according to claim 6, wherein the third search range is set in an area other than the first and second search ranges on the reference screen, and the third search range A motion vector detecting method, wherein the search for the position of the prediction block in the range is performed without referring to the first and second motion compensation errors. 8. The motion vector detection method according to claim 6, wherein, in the first search range, a search process in units of pixels is performed as a process of searching for a position of the prediction block. In the motion vector detection method, the process of searching for the position of the prediction block is performed with an accuracy of half a pixel or more by interpolating pixel values corresponding to virtual pixels between pixels on the reference screen. 9. The motion vector detection method according to claim 6, wherein in the second search range, the position of the prediction block is:
The search is performed along only the vertical or horizontal direction on the reference screen. In the third search range, the position of the prediction block is searched along a direction other than the search direction in the second search range. Characteristic motion vector detection method. 10. The position of a prediction block corresponding to a processing target block to be processed in a reference screen is fixed for each block of a predetermined number of pixels, which is a unit for processing image data corresponding to the processing target screen. A motion vector detection device that performs a motion vector detection process of searching under the condition and detecting information indicating the position of the prediction block as a motion vector corresponding to the block to be processed. Error calculation for calculating an error of image data between a block corresponding to a motion vector search center, which is a start position, and a processing target block as a motion compensation error corresponding to the motion vector search center. Means, a neighboring block located near the block to be processed,
Comparing means for comparing a motion compensation error corresponding to a neighboring block, which is an error of image data with the corresponding prediction block, with a motion compensation error corresponding to the motion vector search center; Determining means for determining whether or not to perform a motion vector detection process on the block to be processed in accordance with the comparison result of the above. Only when the determination means determines that the motion vector detection process should be performed, the motion vector A motion detection means for detecting the motion vector by performing the motion vector detection processing with the search center as a search start position; and detecting by the motion detection means when the determination means determines that the motion vector detection processing should be performed. The detected motion vector is output as a motion vector corresponding to the block to be processed. When it is determined that the motion vector should not be performed, a selection unit that outputs the position information indicating the motion vector search center as a motion vector corresponding to the block to be processed, and a motion corresponding to the motion vector output from the selection unit A memory for temporarily storing the compensation error as data for determining whether or not to perform a motion vector detection process on another block. 11. For each block of a predetermined number of pixels, which is a unit for processing image data corresponding to a screen to be processed, the position of a prediction block corresponding to the block to be processed on the reference screen is fixed. A motion vector detecting device that searches under the condition and detects information indicating the position of the predicted block as a motion vector corresponding to the block to be processed, wherein the motion vector detecting device includes a first search range set on the reference screen; To search for the position of the prediction block, and determine the position of the image data between the block on the reference screen specified by the first candidate motion vector corresponding to the processing target block and the processing target block on the processing target screen. First motion detecting means for calculating a first motion compensation error, which is an error, and a second motion detecting means for calculating a second motion compensation error, the second motion vector being located near a pixel position indicated by the first candidate motion vector. In the search range, the position of the prediction block is searched, and the error of the image data between the block on the reference screen specified by the second candidate motion vector corresponding to the block to be processed and the block to be processed. A second motion estimating means for calculating a second motion compensation error, which is a second motion compensation error, and a third search range on the reference screen set based on the first candidate motion vector and the second candidate motion vector. The position of the prediction block is searched for, and the error of the image data between the block on the reference screen specified by the third candidate motion vector corresponding to the processing target block and the processing target block is obtained. A third motion estimating means for calculating the third motion compensation error, and comparing the first, second, and third motion compensation errors, and according to the comparison result, the first, second, third The candidate motion vector Motion vector detecting apparatus characterized by comprising one, and selection means for selecting a motion vector corresponding to the processed block. 12. A data storage medium storing a program, the program causing a computer to perform a process of detecting a motion vector corresponding to a block to be processed by the motion vector detection method according to claim 1. A data storage medium characterized by being a program. 13. A data storage medium storing a program, the program causing a computer to perform a process of detecting a motion vector corresponding to a block to be processed by the motion vector detection method according to claim 6. A data storage medium characterized by being a program.