KR102032793B1 - Method and Apparatus for effective motion vector decision for motion estimation - Google Patents

Method and Apparatus for effective motion vector decision for motion estimation Download PDF

Info

Publication number
KR102032793B1
KR102032793B1 KR1020180065856A KR20180065856A KR102032793B1 KR 102032793 B1 KR102032793 B1 KR 102032793B1 KR 1020180065856 A KR1020180065856 A KR 1020180065856A KR 20180065856 A KR20180065856 A KR 20180065856A KR 102032793 B1 KR102032793 B1 KR 102032793B1
Authority
KR
South Korea
Prior art keywords
search
image
motion information
motion
sub
Prior art date
Application number
KR1020180065856A
Other languages
Korean (ko)
Other versions
KR20180068910A (en
Inventor
김익균
신경선
엄낙웅
정희범
Original Assignee
한국전자통신연구원
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 한국전자통신연구원 filed Critical 한국전자통신연구원
Priority to KR1020180065856A priority Critical patent/KR102032793B1/en
Publication of KR20180068910A publication Critical patent/KR20180068910A/en
Application granted granted Critical
Publication of KR102032793B1 publication Critical patent/KR102032793B1/en

Links

Images

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/503Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
    • H04N19/51Motion estimation or motion compensation
    • H04N19/573Motion compensation with multiple frame prediction using two or more reference frames in a given prediction direction
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/503Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
    • H04N19/51Motion estimation or motion compensation
    • H04N19/533Motion estimation using multistep search, e.g. 2D-log search or one-at-a-time search [OTS]
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/503Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
    • H04N19/51Motion estimation or motion compensation
    • H04N19/56Motion estimation with initialisation of the vector search, e.g. estimating a good candidate to initiate a search
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/59Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving spatial sub-sampling or interpolation, e.g. alteration of picture size or resolution
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/60Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
    • H04N19/61Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding

Abstract

An efficient motion vector extraction method and apparatus therefor are disclosed. A motion vector extraction method comprising determining a search start position in an original image to perform a spiral motion search and determining whether to perform a search in a sub-sampled image during a P-picture search is performed by performing a spiral motion search using a sub-sampled image. In addition, the detection accuracy can be improved by detecting a plurality of motion vector candidates using a subsampling search, which is a new hierarchical spiral motion search method that combines multiple extension schemes.

Description

Efficient motion vector extraction method and apparatus therefor in motion searching {Method and Apparatus for effective motion vector decision for motion estimation}

The present invention relates to an efficient motion vector extraction method and apparatus therefor in motion searching.

In recent years, the digitization of video and its use is rapidly being made. In the case of TV broadcasting, terrestrial waves are digitized, and home appliances such as DVD recorders and digital video cameras are widely used. In addition, the distribution of video on the Internet, the use of video mails, video telephony, and video conferencing by mobile phones are also made.

Video encoding technology is located at the core of digitization of video. Due to the advancement of technology represented by H.264 and the demand for high resolution represented by HDTV, the amount of video encoding processing is increasing day by day. Among them, an increase in the amount of block matching computation, which occupies most of the computation, is a serious problem.

In particular, software encoders can only perform incomplete motion search, which is a major factor in deterioration of image quality. Therefore, there is a demand for the implementation of a highly efficient search algorithm that can satisfactorily reduce both a large amount of computation and a high search accuracy.

With regard to motion search, various search methods have been proposed. However, in an image with easy motion search, a picture quality of about full search can be achieved, but in some images, image quality deterioration of 1 dB or more comes. This is a major factor that impairs image quality stability in the encoder.

For example, spiral motion search, which is mainly used as a search method of a software encoder because of a small amount of computation, causes a large deterioration of the detection degree depending on the image. Minimizing the computation amount by interrupting the search according to a predetermined rule works badly, and when trapped at a local optimum point, the optimum point in a large area cannot be reached.

On the other hand, hierarchical search of the full search category is bad in terms of computational savings, but it is widely used as a search method of a hardware encoder since no deterioration of the extreme detection degree occurs. However, since the individual matching accuracy is lowered, the detection accuracy tends to be lower than that of spiral search in an image where most of the movement is constant.

The motion search refers to a process of finding the most matching position in the reference video for each block in the video to be encoded. As a criterion for the degree of agreement, the difference absolute sum is generally used. If the pixel value of a pixel in the block to be encoded is B, the candidate vector is v in the encoding object video, and the pixel value I cur (r) and the reference image is I ref (r), then the candidate vector in block B The differential absolute sum SAD for v is expressed by equation (1).

Figure 112018056200107-pat00001
(One)

SAD (B, v) is obtained for some vectors, and v, where SAD (B, v) is minimum, is finally used as a motion vector.

SAD is the most efficient as a comparison standard. Nevertheless, the amount of computation is enormous in all searches that evaluate the entire search point in the search range. For this reason, algorithms for reducing computation such as 3-step search, 4-step search and diamond search have been proposed.

In the spiral motion search, a spiral search is performed from an arbitrary point in the image to the circumference, and the search is stopped at a point that satisfies a predetermined predetermined condition. By accurately predicting the search start position, the search can be performed more efficiently. As a method for determining a search start point, a method of referring to a vector obtained by a motion search process in a previous image, a vector obtained from a neighboring block in the current block, or a method of combining the referenced vectors has been proposed. In addition, as a search interruption rule, it is proposed that the sum of difference absolute values be obtained in order, and stop when the value once descends and rises. In this method, the number of search points can be minimized by searching only a specific area of an image. However, if a prediction miss occurs, the deterioration of image quality and the effect of reducing the amount of computation are insufficient.

Hierarchical search is a method of reducing the resolution by creating a sub-sampling image of a video for which motion is to be obtained, thereby reducing the number of search target points and the amount of individual block matching calculations. Pre-search is performed on the lowest resolution image, and the result is searched for a high resolution image. It is called hierarchical search because it gradually increases the resolution and finally obtains a motion vector from the original image. In this method, by lowering the resolution of the sub-sampled image, the number of search target points is lowered, so that the amount of computation is reduced at a constant ratio regardless of the characteristics of the image, and the entire image is searched. On the other hand, in order to drastically reduce the amount of computation, the resolution must be greatly reduced, so the amount of detailed image information of the image is reduced and the search accuracy is lowered. To solve this problem, a method of referring to surrounding block search results, a method of performing filtering processing on a sub-sampled image, or a method of compensating for the loss of image information by extending a template when obtaining a motion Is being proposed.

Combination of techniques such as (1) step search start position, (2) double search range setting, and (3) adaptive stop condition determination for efficient spiral motion search for each original and sub-sampling video Let's do it.

Determination of the search start position sets several candidate vectors, obtains a match degree (SAD) at the position indicated by each candidate vector, and moves the pixel up, down, left, and right, and searches for the position with the highest match (small SAD). Let it be the starting position. At this time, the number of candidates can be increased to bring the search start position closer to the optimum position. However, comparing a large number of candidates at once results in an increase in the computation amount. Thus, rather than comparing all candidates at once, the candidates are divided into several groups for comparison.

First, for each first candidate group, the degree of agreement of the positions indicated by each candidate is obtained. If the coincidence of the most matched position is higher than the threshold value, the position from which the coincidence is obtained is the search start position. If the coincidence of the most matched position is lower than the threshold value, move to the next candidate group and find the coincidence at the position given by the candidate of that group.

In the following, iteratively repeats until the matching of all candidates is finished until candidates with higher matching are shown.

The group classification for the candidate vector for search start position determination is performed by searching the obtained candidate vector by searching the sub-sampled image, searching the original image where the sub-sampling image search has occurred, or searching the original image without sub-sampling image search.

In the search in the sub-sampled image, candidates obtained by using the spatial correlation of the images are divided into three groups to be the first, second, and third groups, and predetermined positions in the image are the fourth group.

In the case of searching in the original image, when searching in the sub-sampling image is performed in advance, the results are set as the first and second groups, and candidates obtained by using spatial correlation are selected using the third group and temporal correlation. Let the obtained candidate be the fourth group, and the predetermined position in the video is the fifth group. If the search is not performed on the sub-sampled image, the candidate obtained by using the spatial correlation is the first group, the candidate obtained by using the temporal correlation is the second group, and the predetermined position in the image is the third group. .

At the time of candidate extraction of each group, if the newly acquired candidate vector has only a difference of less than three pixels in both the X component and the Y component compared to the previously obtained candidate vector, the candidate is not added to the group.

If the interruption condition is not appropriately set for the characteristics of the image, it is interrupted before the optimum solution is obtained, and a significant deterioration in image quality occurs. Or search continues beyond the optimal solution, leading to a significant increase in computation. Therefore, the search range is set to be double around the search start position. In the small search range inside, searching is performed without interruption determination, thereby preventing a significant deterioration in image quality. In the large outside search range, the search is performed while the interruption decision is made. Since this large search range is set smaller than the original search range (full search range), even if the entire search range is searched without interruption, the amount of calculation is only slightly increased. In the case where the image is generally moving in a certain direction, it is likely that the optimal position is out of the predicted position. Therefore, the standard deviation of the motion vector obtained from the motion search is used as an index for determining the size of the search range.

The stop condition adopts the stop condition using the threshold value. This method stops the search when the point that has the difference absolute sum is less than the threshold value. In the search in the sub-sampling image, 13 types of templates are searched per block, but the maximum template (4 block combined template) value is used for the interruption determination.

The threshold value of the search stop is determined based on the sum of difference absolute values when the motion vector that is the basis of the candidate vector used to determine the search start position is determined. If the motions in the video are similar, the minimum difference absolute sum can be regarded as a similar value, and the difference when each candidate vector is determined as an allowable amount to the sum of the difference absolute values when the candidate vector as the search start position is determined. Add the difference between the maximum and minimum of the sum of the absolute values.

Even if the sum of difference absolute values does not fall below the threshold value, the search stops even if there is no update of the vector between two turns of the spiral.

The present invention improves the search accuracy by detecting a plurality of motion vector candidates by subsampling, which is a new hierarchical spiral motion search method combining a spiral motion search and a multiple combination of extended templates while using a subsampling image. An object of the present invention is to provide an efficient motion vector extraction method and apparatus therefor.

Technical problems other than the present invention will be easily understood through the following description.

According to the embodiment of the present invention, the spiral motion search causes a large deterioration of the detection degree depending on the image. Minimizing the computation amount by interrupting the search according to a predetermined rule works badly, and when trapped at a local optimum point, the optimum point in a large area cannot be reached. Hierarchical search in the full search category is poor in terms of computational savings, but it is widely used as a search method of a hardware encoder because no extreme deterioration of detection accuracy occurs. However, in the case where most of the movements are constant, the degree of detection tends to be lower than that of spiral search. The weaknesses of the two representative methods are addressed by a suitable combination of both techniques.

According to an aspect of the invention, the step of determining the search start position in the original image to perform a spiral motion search; And determining whether to perform a search on a sub-sampled image when searching for a P picture.

The determining whether to perform the search may determine whether to perform the search on the sub-sampled image in units of three pictures of a P picture, a B picture, and a B picture.

The determining whether to perform the search may include: calculating standard deviations of the X component and the Y component of the motion vector calculated each time the P picture search is completed; Comparing the calculated standard deviation with a predetermined threshold; And searching for a movement of the sub-sampled image when the standard deviation is smaller than the threshold.

If the standard deviation is greater than the threshold, the method may further include not searching the sub-sampled image until the next B picture is searched.

Here, the sub-sampling image may be an image having a lower resolution of 1/2 of the width and length of the original image.

In addition, the present embodiment may include dividing a 16 * 16 pixel block of the original image into 8 * 8 pixel blocks; Forming 13 types of templates per 16 * 16 pixel block; And using the 13 kinds of motion vectors calculated from the template as candidate vectors when designating a search start position in the original image search.

According to another aspect of the invention, the search performing unit for performing a spiral motion search by determining the search start position in the original image; And a search range determiner configured to determine whether to perform a search on the sub-sampled image during the P picture search.

The search range determiner may include: a standard deviation calculator configured to calculate standard deviations of the X and Y components of the motion vector calculated each time the P picture is searched; And a standard deviation comparison unit configured to compare the calculated standard deviation with a predetermined threshold, and the search performing unit may search for the motion based on the sub-sampled image when the standard deviation is smaller than the threshold.

Other aspects, features, and advantages other than those described above will become apparent from the following drawings, claims, and detailed description of the invention.

Efficient motion vector extraction method and apparatus for motion search according to the present invention utilize a plurality of sub-sampling search methods, which is a new hierarchical spiral motion search method combining spiral motion search and multiple templates in combination with sub-sampling images. There is an effect of improving the search accuracy (accuracy) by detecting the motion vector candidate.

In addition, according to the present invention, there is provided an efficient motion vector extraction method and apparatus for performing a spiral motion search using a sub-sampling image to greatly improve the search accuracy. The search range can be extended without significantly increasing the amount of computation, and the accuracy of the search can be improved by the expansion of the template, and the increase of the computation amount can be achieved by selecting candidate vectors step by step. It is possible to suppress, and by these advantages, the amount of calculation can be suppressed to the same amount as that of the conventional method, and the image quality can be improved.

1 is a block diagram of a motion vector extraction apparatus according to an embodiment of the present invention.
2 is a diagram illustrating switching of image search to which subsampling of a motion vector extracting apparatus according to an embodiment of the present invention is applied.
3 is a diagram illustrating a sub-sampled image and a search template of a motion vector extracting apparatus according to an embodiment of the present invention.
4 is a diagram showing 13 types of templates by combination of 8x8 pixel blocks of the motion vector extracting apparatus according to the embodiment of the present invention;

As the invention allows for various changes and numerous embodiments, particular embodiments will be illustrated in the drawings and described in detail in the written description. However, this is not intended to limit the present invention to specific embodiments, it should be understood to include all changes, equivalents, and substitutes included in the spirit and scope of the present invention.

Terms including ordinal numbers such as first and second may be used to describe various components, but the components are not limited by the terms. The terms are used only for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the first component may be referred to as the second component, and similarly, the second component may also be referred to as the first component. The term and / or includes a combination of a plurality of related items or any item of a plurality of related items.

When a component is referred to as being "connected" or "connected" to another component, it may be directly connected to or connected to that other component, but it may be understood that other components may be present in between. Should be. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. As used herein, the terms "comprise" or "have" are intended to indicate that there is a feature, number, step, action, component, part, or combination thereof described on the specification, and one or more other features. It is to be understood that the present invention does not exclude the possibility of the presence or the addition of numbers, steps, operations, components, components, or a combination thereof.

In addition, the terms “… unit”, “… module”, “… means” described in the specification mean a unit for processing at least one function or operation, which may be implemented by hardware or software or a combination of hardware and software. Can be.

In addition, in the description with reference to the accompanying drawings, the same components regardless of reference numerals will be given the same reference numerals and duplicate description thereof will be omitted. In the following description of the present invention, if it is determined that the detailed description of the related known technology may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted.

1 is a block diagram of a motion vector extraction apparatus according to an embodiment of the present invention. Referring to FIG. 1, a motion vector extracting apparatus 100, a search performing unit 110, a search range determining unit 120, a standard deviation calculating unit 121, a standard deviation comparing unit 122, and a control unit 130 are provided. This is shown.

Background Art The video encoding processing technology has been increasing day by day due to advances in technology represented by H.264 and the demand for higher resolution of a target video represented by HDTV. For this reason, a lot of researches on the amount of computation reduction have conventionally been conducted regarding the motion search which occupies most of the computation quantity. Among them, the spiral motion search has been improved as an efficient method, but it shows a problem such as an extreme decrease in the accuracy of detection in a specific image due to the fact that the actual search is limited to only a part of the search range.

Because of the small amount of computation, spiral motion search, which is mainly used as a search method for software encoders, causes a large deterioration in the detection accuracy depending on the image. Minimizing the computation amount by interrupting the search according to a predetermined rule works badly, and when trapped at a local optimum point, the optimum point in a large area cannot be reached. Hierarchical search in the full search category is poor in terms of computational savings, but it is widely used as a search method of a hardware encoder because no extreme deterioration of detection accuracy occurs. However, in the case where most of the movements are constant, the degree of detection tends to be lower than that of spiral search.

According to this embodiment, the weaknesses of the two representative methods are solved by an appropriate combination of both techniques. Specifically, this problem is solved by extracting a plurality of motion vector candidates by subsampling, which is a new hierarchical spiral motion search method combining spiral motion search and multiple templates using extended sampling while using subsampling images. By applying this proposal, we can improve the accuracy of the search while keeping the computational level at the same level as the existing fast search method.

The search performer 110 determines a search start position in the original image and performs a spiral motion search. The search range determiner 120 determines whether to perform a search on a sub-sampled image when searching a predetermined region, for example, a P picture.

The standard deviation calculator 121 calculates the standard deviation of the X and Y components of the calculated motion vector each time the P picture search is completed, and the standard deviation comparator 122 compares the calculated standard deviation with a predetermined threshold. do. The control unit 130 controls the search execution unit 110, the search range determination unit 120, the standard deviation calculator 121, and the standard deviation comparison unit 122, respectively.

According to the present embodiment, the spiral motion search performance can be improved by using the sub-sampling image. The spiral motion search searches only a specific area of an image and can minimize the search score. However, there is a problem that if the search start position prediction fails, the image quality is greatly deteriorated, and if it is wrong, the computation amount is greatly increased.

In order to solve this problem, if the setting is made difficult to interrupt, and the range of the search is widened, the amount of calculation is greatly increased and the benefits of the spiral motion search are lost.

Therefore, as a hierarchical search method aiming to expand the search range without increasing the amount of computation, a spiral search that adaptively switches by P picture period units whether or not to perform sub-sampling image layer search is proposed.

In order to deteriorate the search accuracy in the sub-sampling video layer, which is a problem in hierarchical motion search, a plurality of extended templates are used together and other block motion detection results that correlate temporally and spatially are actively used as candidates for search start position. Try to solve it.

The switching of the sub-sampling image search is as follows. The search in the sub-sampled video is performed by searching twice with the original video, which leads to an increase in the amount of computation. In an image that is easy to search, it is not necessary to search in the sub-sampling image when sufficient image quality is maintained even in the conventional technique.

Therefore, it is determined whether or not the search is performed in the sub-sampled video by three picture units of P, B, and B based on the characteristics of the video, and adaptively switches the use of the search. Each time the P picture search is completed, the standard deviation of the X and Y components of the obtained motion vector is obtained.

As the proposed technique is based on the spiral motion search, the motions of neighboring blocks and temporal blocks are referred to as predictive vectors. At this time, the small deviation means that only similar predictive vectors are acquired constantly, and if a different motion is included in part, it cannot correspond to the motion. Conversely, when the standard deviation increases, a predictive vector is obtained in various directions, and a plurality of locations widely distributed can be searched for. From this fact, the proposed technique uses a sub-sampled image to search a wide range when the standard deviation is small. When the sum of the standard deviations is equal to or larger than the threshold value, the search in the sub-sampled picture at the time of searching for the next P and the next B picture is omitted. The threshold value was 40 obtained experimentally.

According to FIG. 2, in step S210, a search is performed on a P picture, in step S220, a standard deviation for motion is calculated, and in step S230, the standard deviation is compared with a threshold. As a result of the comparison, when the standard deviation is larger than the threshold, the search is performed on the original image only in step S240. If the standard deviation is smaller than the threshold, the search is performed on the sub-sampled image and the original image. Here, if the standard deviation is larger than the threshold, the sub-sampled image may not be searched until the next B picture search for the next P picture.

According to this embodiment, the setting of the sub-sampled image is as follows. As described above, if the resolution is reduced, the search point can be reduced and the computation amount can be reduced. However, if the resolution is greatly reduced, the detail of the image is broken and the search accuracy is lowered. In general, hierarchical search reduces the resolution to a maximum of 1/4 in width and width, but in this embodiment based on spiral search, only a part of the search range is different from the conventional hierarchical search. Therefore, since the amount of computational reduction is large compared with the general hierarchical search, it is not necessary to greatly reduce the resolution. Therefore, the sub-sampled image to be searched is an image having a width of 1/2 and a resolution of 1/4 as a whole with respect to the original image.

According to the present embodiment, a template may be extended in the sub-sampling image search. Regarding the motion search in the sub-sampled image, lowering the resolution leads to a decrease in the search accuracy. Therefore, the template when comparing blocks is extended. By expanding the template, the pixels around the original motion detection target block are compared, and as a result, the search accuracy is improved.

Simply expanding the template around will result in overlapping comparison operations and increased computation. Therefore, as shown in FIG. 3, the sub-sampled image b having the horizontal and vertical half of the original image a having the 16 x 16 pixel block 310 and having the 8 x 8 pixel block 320 is present. ), The sub-sampled image c is divided into 16 x 16 pixel blocks 330, and motion is calculated in units of blocks. As a result, motion detection for four blocks of the original image is simultaneously performed. As a result, template expansion is realized without duplication of comparison operations.

When the extended template spans the boundary between the objects of different movements in the image, the search accuracy is lowered. Therefore, a plurality of templates having different directions of expansion are provided so that no one can cross the boundary. Specifically, the 16 x 16 pixel block 330 is divided into four 8 x 8 pixel blocks 340 which are original sizes, and by this combination, 13 types of templates as shown in FIG. 4 are formed per block. do.

(A) of FIG. 4 is a template of 1 block, (b) 2 blocks, (c) 3 blocks, and (d) 4 blocks. In comparison, the movements with the highest agreement are detected in these templates. The 13 kinds of motion vectors obtained from these results are used as candidate vectors for designating a search start position when searching in the original image.

In addition, a detailed device configuration diagram of a motion vector extracting apparatus according to an embodiment of the present invention, a common platform technology such as an embedded system, an O / S, a communication protocol, an interface standardization technology such as an I / O interface, and the like will be described. As it is obvious to those skilled in the art, it will be omitted.

The motion vector extraction method according to the present invention can be implemented in the form of program instructions that can be executed by various computer means and recorded in a computer readable medium. In other words, the recording medium may be a computer-readable recording medium having recorded thereon a program for causing the computer to execute the above-described steps.

As described above, each component and / or function described in each embodiment may be implemented in combination with each other, and those skilled in the art to which the present invention described in the claims below It will be understood that various modifications and variations can be made in the present invention without departing from the spirit and scope.

100: motion vector extraction device
110: search performing unit
120: search range determination unit
121: standard deviation calculation unit
122: standard deviation comparison unit

Claims (5)

  1. An image encoding method of generating a bitstream by encoding an encoding target image.
    Extracting a plurality of motion information candidates for a target block included in the encoding target image; And
    Encoding motion information of the target block based on the plurality of motion information candidates,
    Extracting the plurality of motion information candidates,
    Calculating a first motion information component and a second motion information component for the target block;
    Identifying a difference based on the first motion information component and the second motion information component; And
    And extracting motion information candidates of the target block based on the verification result.
  2. The method of claim 1,
    Any one of the plurality of motion information candidates is derived by searching in a sub-sampling image having a lower resolution than the original image.
  3. delete
  4. The method of claim 2,
    And the sub-sampling image is an image whose resolution is reduced to 1/2 of the width and length of the original image.
  5. Extracting a plurality of motion information candidates for the target block included in the encoding target image; And
    Encoding motion information of the target block based on the plurality of motion information candidates,
    Extracting the plurality of motion information candidates,
    Calculating a first motion information component and a second motion information component for the target block;
    Identifying a difference based on the first motion information component and the second motion information component; And
    And extracting a motion information candidate of the target block based on the verification result.
KR1020180065856A 2018-06-08 2018-06-08 Method and Apparatus for effective motion vector decision for motion estimation KR102032793B1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
KR1020180065856A KR102032793B1 (en) 2018-06-08 2018-06-08 Method and Apparatus for effective motion vector decision for motion estimation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
KR1020180065856A KR102032793B1 (en) 2018-06-08 2018-06-08 Method and Apparatus for effective motion vector decision for motion estimation

Publications (2)

Publication Number Publication Date
KR20180068910A KR20180068910A (en) 2018-06-22
KR102032793B1 true KR102032793B1 (en) 2019-10-17

Family

ID=62768650

Family Applications (1)

Application Number Title Priority Date Filing Date
KR1020180065856A KR102032793B1 (en) 2018-06-08 2018-06-08 Method and Apparatus for effective motion vector decision for motion estimation

Country Status (1)

Country Link
KR (1) KR102032793B1 (en)

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100772405B1 (en) * 2005-09-27 2007-11-01 삼성전자주식회사 Methods for adaptive noise reduction based on global motion estimation and video processing system therefore
KR101443701B1 (en) * 2010-11-22 2014-09-29 한국전자통신연구원 Adaptive motion search range determining apparatus and method

Also Published As

Publication number Publication date
KR20180068910A (en) 2018-06-22

Similar Documents

Publication Publication Date Title
KR101023263B1 (en) Predictive reference information generation method, dynamic image encoding and decoding method, device thereof, and storage medium containing the program thereof
US6418166B1 (en) Motion estimation and block matching pattern
RU2381630C2 (en) Method and device for determining block conformity quality
US6567469B1 (en) Motion estimation algorithm suitable for H.261 videoconferencing applications
US6380986B1 (en) Motion vector search method and apparatus
US20050238103A1 (en) Method and apparatus for performing high quality fast predictive motion search
JP5281891B2 (en) Adaptive motion search range
US8705611B2 (en) Image prediction encoding device, image prediction encoding method, image prediction encoding program, image prediction decoding device, image prediction decoding method, and image prediction decoding program
JP3847827B2 (en) Motion vector detection method
JP2007202150A (en) Method of and apparatus for deciding encoding mode for variable block size motion estimation
JP2010517416A (en) Motion estimation for discovered frame regions
JP4004653B2 (en) Motion vector detection method and apparatus, and recording medium
US20030189980A1 (en) Method and apparatus for motion estimation between video frames
US6549576B1 (en) Motion vector detecting method and apparatus
US8064522B2 (en) Motion-vector detecting device, motion-vector detecting method, and computer program
JP2004522340A (en) Local constraints for motion matching
US7720154B2 (en) System and method for fast variable-size motion estimation
US6842483B1 (en) Device, method and digital video encoder for block-matching motion estimation
WO2011095260A1 (en) Managing predicted motion vector candidates
JP5044568B2 (en) Motion estimation using predictive guided decimation search
US8009923B2 (en) Method and system for motion estimation with multiple vector candidates
JP4555758B2 (en) Coding mode selection method for intra prediction in video compression
JP2003274416A (en) Adaptive motion estimation apparatus and method
JP2004530367A (en) Motion vector prediction method and motion vector prediction device
US20110228853A1 (en) Moving image encoding and decoding system

Legal Events

Date Code Title Description
A201 Request for examination
A107 Divisional application of patent
E902 Notification of reason for refusal
E902 Notification of reason for refusal
E701 Decision to grant or registration of patent right
A107 Divisional application of patent