TWI224930B - Calculation method for prediction motion vector - Google Patents

Calculation method for prediction motion vector Download PDF

Info

Publication number
TWI224930B
TWI224930B TW92100445A TW92100445A TWI224930B TW I224930 B TWI224930 B TW I224930B TW 92100445 A TW92100445 A TW 92100445A TW 92100445 A TW92100445 A TW 92100445A TW I224930 B TWI224930 B TW I224930B
Authority
TW
Taiwan
Prior art keywords
motion vector
block
reference image
motion vectors
same
Prior art date
Application number
TW92100445A
Other languages
Chinese (zh)
Other versions
TW200402238A (en
Inventor
Byeong-Moon Jeon
Yoon-Seong Soh
Original Assignee
Lg Electronics Inc
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 Lg Electronics Inc filed Critical Lg Electronics Inc
Publication of TW200402238A publication Critical patent/TW200402238A/en
Application granted granted Critical
Publication of TWI224930B publication Critical patent/TWI224930B/en

Links

Landscapes

  • Compression Or Coding Systems Of Tv Signals (AREA)

Abstract

A method for calculating a prediction motion vector of a block to be coded, by performing a median operation using motion vectors of neighboring blocks. If one of the neighboring blocks has motion vectors MV1 and MV2, direction (forward or backward) information of reference pictures referred to, respectively, by the motion vectors are acquired and one of the motion vectors is selected with reference to the acquired direction information. Then, the median operation including the selected motion vector is performed to obtain the prediction motion vector of the block to be coded.

Description

1224930 案號 921004451224930 Case No. 92100445

五、發明說明(1) [發明所屬之技術領域] 本發明有關一種動晝編竭系統,尤其有關一 動向量計算方法。其中係於欲編碼區塊環周之鄰區 具有多數運動向量時,定義欲在中值運算中使用的 〜 之預 量,以使用鄰區塊之運動向量資訊,獲得欲編碼心動向 測運動向量(PMV),並改進編碼效率 4 [先前技術] 通常,為了減少傳达運動資訊所須使用的位元 碼器並不直接將運動向量MV傳送給解碼器,而是經由鵠 運算選取三個鄰區塊之運動向量的中間值、決定選中值 間值作為預測運動向量PMV、獲得Mv與^¥之間的差^中 (亦即,MVD = MV-PMV)、並將獲得的差值MVD傳送給 H V % 旦奴後,解碼器以相同於編碼器的方式獲得預測運 里PMV,並相加傳送的MVD與所獲得的預測運動向量勒向 以獲得運動向量。圖,區塊£是欲編碼(或解碼)之’ 塊,區塊A、B、c則為區塊£之鄰區塊。將鄰區塊A、B品 之運動向量分別定義為1«^、MVB、MVC,則可經由如、C 中值運算,獲得區塊E之預測運動向量: 之 PMV=中間值{mva、MVB、MVC} 田區塊C在圖像之外時,則使用圖1中的區塊d替 品塊C一。倘若三個區塊A、B、C或A、B、D中,只有一個區 鬼表示,基準圖像與區塊E表示的一樣,則使用該區塊的 運動向lMV作為預測運動向量PMV。所有圖像,不論其類V. Description of the invention (1) [Technical field to which the invention belongs] The present invention relates to a moving day exhaustion system, and more particularly to a method for calculating a motion vector. Among them, when the neighboring area of the block to be coded has a majority of motion vectors, the pre-value of ~ to be used in the median operation is defined to use the motion vector information of the neighboring block to obtain the heartbeat motion vector (PMV) ) And improve coding efficiency 4 [Prior art] Generally, in order to reduce the number of bit encoders necessary to communicate motion information, the motion vector MV is not directly transmitted to the decoder, but three neighboring blocks are selected through unitary operations. The middle value of the motion vector, determine the selected value as the predicted motion vector PMV, obtain the difference between Mv and ^ ¥ ^ (that is, MVD = MV-PMV), and transmit the obtained difference MVD to After HV%, the decoder obtains the predicted PMV in the same way as the encoder, and adds the transmitted MVD and the obtained predicted motion vector to obtain a motion vector. In the figure, block £ is the block to be encoded (or decoded), and blocks A, B, and c are neighboring blocks of block £. If the motion vectors of the neighboring blocks A and B are defined as 1 «^, MVB, and MVC, the predicted motion vectors of block E can be obtained through median operations such as C: PMV = median {mva, MVB , MVC} When block C is outside the image, block C in block d in Figure 1 is used instead. If only one of the three blocks A, B, C or A, B, and D is represented by ghosts, and the reference image is the same as that of block E, the motion direction lMV of the block is used as the predicted motion vector PMV. All images, regardless of their type

第5頁 1224930 _塞號 92100445_年月曰_修正_ 五、發明說明(2) 型如何,皆適用此種運動資訊傳送方法。 另一方面,B圖像具有五種預測模式,諸如正向模 式、反向模式、雙預測模式、直接模式、及在内(inti;a) 模式。通常,正向模式之鄰區塊其運動向量MVFW是從索引 為ref 一 idx — fWd的正向基準圖像獲得的,而反向模式之鄰 區塊其運動向量MVBW是從索引為ref — idx 一 bwd的反向基準 圖像獲得的。 & 在B圖像之雙預測模式中,允許不同方向與相同方向 的預測,諸如正向/正向、反向/反向、及正向/反向。每 一基準圖像均使用索引ref— idx一 fwd或ref 一idx一bwd,不 論其方向如何(正向或反向),且每一運動向量也都用訂FW 或MVBW表示,不論其方向如何。(其理由在於,預定義 之’語法’都依其原有方式使用。關於語法之陳述,每一索 引可使用’ref_idx—10’或’ref一idx —11’ ,而每一運動向量 則可使用’鮮一1131:0’或’111¥一1131:1’。)6圖像之直接模式是 一種預測模式,其中並不將運動資訊傳送到解碼器,運動 向量MVf、MVb及基準圖像是從解碼器内部導出的。事實 上,所導出的運動向量都以MVf、M Vb表示,不論其方向如 何,此點係與雙預測模式中的情況相同。 在習用的B圖像預測運動向量PWV計算方法中,僅取各 鄰區塊之正向運動向量’並針對所取的正向運動向量執行 中值運算,以獲得區塊E之正向預測運動向量。若鄰區塊 之一沒有正向運動向量’則將其運動向量設定為0,並在 此一情況下執行中值運算。此種方法同樣應用在區塊E之 反向預測運動向量’以便僅使用鄰區塊之反向運動向量。Page 5 1224930 _ plug No. 92100445 _ month month _ correction _ V. Description of the invention (2) Whatever type is used, this kind of sports information transmission method is applicable. On the other hand, the B picture has five prediction modes, such as a forward mode, a reverse mode, a bi-prediction mode, a direct mode, and an inti (a) mode. Generally, the motion vector MVFW of the neighboring blocks in the forward mode is obtained from the forward reference image with the index ref_idx — fWd, and the motion vector MVBW of the neighboring blocks in the reverse mode is from the index ref — idx. A bwd inverse reference image was obtained. & In the B-picture bi-prediction mode, predictions in different directions and the same direction are allowed, such as forward / forward, reverse / reverse, and forward / reverse. Each reference image uses the index ref_idx_fwd or ref_idx_bwd, regardless of its direction (forward or reverse), and each motion vector is also represented by a custom FW or MVBW, regardless of its direction . (The reason is that the predefined 'grammars' are used in their original way. Regarding the statement of syntax, each index can use 'ref_idx-10' or 'ref_idx-11', and each motion vector can use 'Xianyi 1131: 0' or '111 ¥ 一 1131: 1'.) The direct mode of 6 images is a prediction mode in which motion information is not transmitted to the decoder. The motion vectors MVf, MVb and the reference image are Exported from inside the decoder. In fact, the derived motion vectors are all represented by MVf and M Vb. Regardless of their direction, this point is the same as that in the bi-prediction mode. In the conventional B-picture prediction motion vector PWV calculation method, only the forward motion vectors of neighboring blocks are taken and a median operation is performed on the obtained forward motion vectors to obtain the forward prediction motion of block E. vector. If one of the neighboring blocks does not have a forward motion vector ', its motion vector is set to 0, and a median operation is performed in this case. This method is also applied to the backward prediction motion vector 'of block E so as to use only the backward motion vectors of neighboring blocks.

第6頁 1224930Page 6 1224930

案號 921004^ 五、發明說明(3) 若鄰區塊之一為,在内,模式,則其運動向量設定為〇,並 將該鄰區塊表示的基準圖像,視為異於區塊E表示”''的基準 圖像,並在此一情況下獲得預測運動向量PMV。 然而,如上所述,在B圖像之雙預測模式中,允許從 不同方向與相同方向預測’諸如正向/正向、反向/反向、 及正向/反向,每一基準圖像均使用索引ref 一 idx 一 fwd或 ref —idx 一 bwd,不論其方向如何(正向或反向),且每一運 動向量也都用MVFW或MVBW表示,不論其方向如何。因此, 必須定義一種方法,在某一鄰區塊有兩個運動向量存在 時,用以計算預測運動向量PMV。 假若某一鄰區塊為雙預測模式(或直接模式)時,立運動向 量MVFW、MVBW(或MVf、MVb)可為相同方向(諸如正向/正向 或反向/反向),或為不同方向(諸如正向/反向)。此種運 動向量之方向資sfl無法單從運動向量語,mvfw,及,^心⑺,或 基準圖像索引’ref 一 idx —fwd,及,ref —idx—bwd,判定。習用 之B圖像PMV計算方法不能提供此種問題的精確說明,因此 導致極大的困擾。 例如,當某一鄰區塊為雙預測模式而具有方向為正向/正 向的二個運動向量時,習用之PMV計算方法對於計算區塊£ 之正向預測運動向量PMV時,是否必須同時使用二個或可 僅使用其中任一個運動向量,並未提供清晰定義的決定。 [發明内容]Case No. 921004 ^ V. Description of the invention (3) If one of the neighboring blocks is in the inner mode, its motion vector is set to 0, and the reference image represented by the neighboring block is regarded as a different block. E represents the reference image of "", and in this case the predicted motion vector PMV is obtained. However, as described above, in the bi-prediction mode of the B image, it is allowed to predict 'such as forward from different directions and the same direction' / Forward, reverse / reverse, and forward / reverse, each reference image uses the index ref-idx-fwd or ref —idx-bwd, regardless of its direction (forward or reverse), and Each motion vector is also represented by MVFW or MVBW, regardless of its direction. Therefore, a method must be defined to calculate the predicted motion vector PMV when two motion vectors exist in a neighboring block. If a neighbor When the block is in bi-prediction mode (or direct mode), the vertical motion vectors MVFW, MVBW (or MVf, MVb) can be in the same direction (such as forward / forward or reverse / reverse), or different directions (such as Forward / reverse). The direction of this motion vector cannot be sfl Judging from the motion vector language, mvfw, and ^ heart palpitations, or the reference image index 'ref_idx_fwd, and ref_idx_bwd, the conventional B-picture PMV calculation method cannot provide the accuracy of such problems. This leads to great confusion. For example, when a neighboring block is a bi-prediction mode and has two motion vectors in the forward / forward direction, the conventional PMV calculation method is used to calculate the forward prediction of the block £. In the case of PMV, whether two of them must be used at the same time or only one of them can be used, does not provide a clearly defined decision. [Summary of the Invention]

第7頁 案號 92100445 五、發明說明(4) 因此,鑑於上述問題,特 於提供一種對基準圖像指 基準圖像方向之方法,其 ,運動向量方向資訊的唯 每一基準圖像的方向資訊 本發明另一目的在於提供 中係於一欲編碼區塊之鄰 ft值運算中使用的運動 直資訊,獲得欲編碼區塊 蝙碼效率。 根據本發明之一層面,藉 為該基準圖像之特二 ϊ::τ向資訊,可 季乂佳是,可用一 g! 準pm 圖像順序 旱圖像顯不順序之方向次 根據本發明另一屑 貝 :圖像索引分別^的ί 的步驟有:取得每—義 之顯示順序資訊與曰^ / 每一基準圖像相對目^欲 較佳是,每一基準=欲 數(p〇c)值獲得。圖像之 根據本發明再一層面 執行中值運算的:蝙= 曰 此提出本發明。本發明之目的在 定方向資訊之方法,盥一播冰 中對每一基準圖像指^可幫^獲 一資汛,以獲得從每一鄰區塊至 :種預測運動向量計算方法,其 區塊有多數運動向量時,定義欲 向量,以便利用鄰區塊之運動向 之預測運動向量(PM V ),並改進 由指定一基準圖像之方向資訊作 對基準圖像索引指標的基準圖像 述及其他目的。 計數(POC )值表示可指示每一基 訊。 中和:供一種方法’用以決定以夷 個基準圖像的方向。此方法包括 圖像的顯示順序資訊,比較取得 編碼區塊之顯示順序,以及決二 編碼區塊之方向(正向或反向) 顯示順序資訊可從一圖像順序古十 中提供一種使用鄰區塊運動向量 塊預測運動向量(PMV)計算方里Case No. 92100445 on page 7 V. Description of the invention (4) Therefore, in view of the above problems, a method for referring to a reference image with respect to a reference image is provided. The motion vector direction information is unique to each reference image's direction. Information Another object of the present invention is to provide motion straight information used in the adjacent ft value calculation of a block to be coded to obtain the bat efficiency of the block to be coded. According to one aspect of the present invention, by using the special ϊ :: τ direction information of the reference image, it may be better to use a g! Quasi-pm image order and the order of the dry image display order according to the present invention. Another step: the steps of the image index are: obtaining the display order information of each meaning and the meaning of each relative image ^ / For each reference image is better, each reference = desire number (p〇c ) Value obtained. Image According to another aspect of the present invention, the median operation is performed: bat = said The present invention is proposed. The purpose of the present invention is to determine the direction information. Using a reference image for each reference image in a sowing ice can help to obtain a flood, from each neighboring block to: a method for calculating a predicted motion vector, which When the block has most motion vectors, define the desired vector so as to use the predicted motion vector (PM V) of the motion direction of the neighboring block and improve the reference image by specifying the direction information of a reference image as the reference image index index Address other purposes. The count (POC) value indicates that each base can be indicated. Neutralization: For a method 'to determine the orientation of a reference image. This method includes the display order information of the images, comparing and obtaining the display order of the coded blocks, and the direction (forward or reverse) of the second coded block. The display order information can provide a use of neighbors from an image order. Block motion vector block prediction motion vector (PMV) calculation method

第8頁 1224930 案號 92100445 曰 修正 五、發明說明(5) 法,其步驟包括:a )若該 得以鄰區塊運動向量指向 照取得的方向資訊來選取 包括所選取運動向量之中 測運動向量。 較佳是,以上步驟a)可包 量指向之基準圖像之顯示 序資訊,以決定各運動向 較佳是,若該等鄰區塊之 時,以上步驟b)可包括以 方向與預測運動向量方向 向量之中值運算,以獲得 或者,若該等鄰區塊之一 該運動向量之方向與預測 驟b)可包括以下步驟:將 一鄰區塊表示的基準圖像 像;並執行包括零運動之 量。 或者,若該等鄰區塊之一 向相同並與預測運動向量 MV1、MV2都表示同一基準 步驟b-1):選取二個運動 所選取運動向量之中值運 更佳是,以上步驟b-Ι)可 等鄰區塊具有運動向量時,先取 的基準圖像的方向資訊;及b)參 各鄰區塊運動向量之一,並執行 值運算,以獲得欲編碼區塊之預 括以下步驟 順序資訊與 量之方向資 一有二個不 下步驟:從 相同者;並 預測運動向 有二個相同 運動向量之 該二運動向 異於欲編碼 中值運算, :比較 欲編碼 訊。 同方向 二個運 執行包 量。 方向之 方向不 量設定 區塊表 以獲得 鄰區塊運動向 區塊之顯不順 之運動向量 動向量中選取 括所選取運動 運動向量,但 同時,以上步 為0 ;假定該 示的基準圖 預測運動向 有二個運動向量MV1、MV2,其方 之方向相同,且二個運動向量 圖像時,以上步驟b )可包括以下 向量MV1、MV2之一,並執行包括 算,以獲得預測運動向量。 包括以下步驟:從二個運動向量Page 8 1224930 Case No. 92100445 Amendment V. Invention Description (5) The method includes the following steps: a) if the neighboring block motion vector points to the direction information obtained according to the selected motion vector, the medium motion vector is selected and included. . Preferably, the above step a) may include the display order information of the reference image pointed to to determine the movement direction. Preferably, if the neighboring blocks are used, the above step b) may include the direction and the predicted movement. Vector direction vector median operation to obtain or, if one of the neighboring blocks, the direction and prediction step of the motion vector b) may include the following steps: a reference image image represented by a neighboring block; Zero amount of motion. Alternatively, if one of the neighboring blocks is the same and represents the same reference step b-1 as the predicted motion vectors MV1 and MV2: the median value of the selected motion vector selected by the two motions is better, step b-1 above. ) When the neighboring blocks have motion vectors, the orientation information of the reference image is taken first; and b) one of the neighboring block motion vectors is referenced, and a value operation is performed to obtain the following sequence of steps for the block to be encoded There are two steps in the direction of information and quantity: from the same one; and predicting the direction of movement of the two movements with two identical motion vectors is different from the median operation of encoding, comparing the information to be encoded. Two shipments in the same direction. Set the block table to determine the direction of the direction. The selected motion vector is selected from the motion vector of the adjacent block to the block. The motion vector is selected, but at the same time, the previous step is 0. Assume that the shown reference map predicts In the motion direction, there are two motion vectors MV1 and MV2, the directions of which are the same, and when there are two motion vector images, the above step b) may include one of the following vectors MV1 and MV2, and perform an include operation to obtain the predicted motion vector. . Includes the following steps: from two motion vectors

第9頁 1224930 修正 Μ 曰 案號 92100445 五、發明說明(6) MV1、MV2中,選取欲較先解碼者,或選取模式(MV1模 MV2模式)與預測運動向量之模式相同者,並執行包括 取運動向量之中值運算,以獲得預測運動向量。其中^ = 一具有相同模式之運動向量,表示其轉錄係與指示 = 動向量之轉錄相同。 j運 或者,若該等鄰區塊之一有二個運動向量_、MV2 向相同並與預測運動向量之方向相同,但二個:方 時,以上步驟b)可包括以下所表示的基準圖像 M:2 ,,€ ^ , , , . , „ ^ ^ ' ;動;;行包括所選取運動向量之中值運算,以獲得預剛 若該等鄰區塊之-有二個運動 向相同並與預測運動向量之方向 fMviy y;其方 MVl、MV2都非表示欲編碼區塊:沾但一個運動向量 示不同的基準圈後時,以表不的基準圖像’而且表 運動向量MV1、MV2中,其表干夕a、、隹括以下乂驟·選取 塊所表示基準 ^_ 土準圖像最接近欲編碼區 y圖像者’ϋ執行包括所選取…玫 獲得預測運動向量。 里4 γ徂逆异,以 或者,若該等鄰區塊之一有一彳 量方向相異時,以上步驟b)可包=量係與預測運動向 德W Μ Π ,Α 疋邊郇區塊表不的基準同 、扁碼區塊表示的基準圖像;並執行包括該—鄰區Page 9 1224930 Amendment No. 92100445 V. Description of the Invention (6) Among MV1 and MV2, select the one to be decoded first, or select the same mode (MV1 mode and MV2 mode) as the mode that predicts the motion vector, and execute the include Take the median of the motion vector to obtain the predicted motion vector. Where ^ = a motion vector with the same pattern, indicating that its transcription is the same as the transcription of the instruction = motion vector. Or, if one of the neighboring blocks has two motion vectors _, MV2 are the same and the direction of the predicted motion vector is the same, but two: square time, the above step b) may include the reference map shown below Like M: 2,, € ^,,,., „^ ^ '; Move; the line includes the median operation of the selected motion vector to obtain the pre-rigid if two neighboring blocks have the same motion direction And the direction fMviy y of the predicted motion vector; neither MV1, MV2 represents the block to be coded: when a motion vector shows a different reference circle, the reference image is represented, and the motion vector MV1, In MV2, the following steps are used to describe the following steps: • Select the reference represented by the block ^ _ The standard image is closest to the image to be coded in the area y. The implementation includes the selected ... to obtain the predicted motion vector. 4 γ 徂 is inverse, or if one of the neighboring blocks has a different volume direction, the above step b) may include the volume system and the predicted motion direction W Μ Π, Α 疋 side block table Not the same as the reference image represented by the flat code block;

第10頁 1224930Page 10 1224930

塊零運動之中值運算,以獲 或者,若該等鄰區塊之一有 量方向相同時,以上步驟b) 一鄰區塊該一運動向量之中 量。 得預測運動向量。 一個運動向量係與預測運動向 可包括以下步驟:執行包括該 值運异’以獲得預測運動向 可更加清晰了解本發明上 藉由以下參照附圖之詳細說明 述及其他目的、特徵與優點。 [實施方式]Block zero motion median operation to obtain or if one of the neighboring blocks has the same quantity direction, step b) above, the motion vector of a neighboring block. Have to predict the motion vector. A motion vector system and the predicted motion direction may include the following steps: Performing the operation including the value to obtain the predicted motion direction may provide a clearer understanding of the present invention. The other objects, features, and advantages are described in the following detailed description with reference to the accompanying drawings. [Embodiment]

=一欲編碼d塊之$ _鄰1塊為雙預測模式(或直接模式) ^ '、運動向1MVFW與MVBW(或mvf與mvb)可能為相同方 向,諸如正向/正向及反向/反向,或為不同方向,諸如正 向/反向。此種運動向量的方向資訊無法單從運 法1VFW’與,MVBW,或基準圖像索 μ 引ref —idx —fwd’與’ ref一idx —bwd,判定。因此,必須參照 ,基準圖像所保有的各不相同的唯一資訊,以獲得方向資 訊0 、 本發明提供一種獲得運動向量方向資訊的方法,其中包 括··比較各基準圖像之顯示順序;以及,以獲得的方向資 訊為基礎’計算一預測運動向量PMV。 現在參照附圖詳細說明本發明各較佳實施例。 嫌 1 · ^經由比較顯示順序,確認運動向量方向 ^异欲編碼區塊之預測運動向量PM V前,必須先獲得各鄰 區塊運動向量的方向資訊。然後,經由各鄰區塊運動向量 、 的方向資訊,決定中值運算中是否必須包括各鄰區塊之運= $ To be coded for d-blocks. _ Adjacent 1 block is a bi-prediction mode (or direct mode) ^ ', the motion direction 1MVFW and MVBW (or mvf and mvb) may be the same direction, such as forward / forward and reverse / Reverse, or a different direction, such as forward / reverse. Such direction information of the motion vector cannot be determined solely from the methods 1VFW ′ and MVBW, or the reference image index μ ref —idx —fwd ’and‘ ref —idx —bwd. Therefore, it is necessary to refer to the different unique information held by the reference image to obtain the direction information 0. The present invention provides a method for obtaining the direction information of the motion vector, which includes comparing the display order of each reference image; and To calculate a predicted motion vector PMV based on the obtained direction information. Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings. 1 ^ Confirm the direction of the motion vector by comparing the display order. ^ Before predicting the motion vector PM V of the heterosexual coded block, you must first obtain the direction information of the motion vector of each neighboring block. Then, based on the direction information of the motion vectors of the neighboring blocks, determine whether the operation of each neighboring block must be included in the median operation.

第11頁Page 11

案號 92100445 五、發明說明(8) 動向量。 若;ΐ 具::個運動向量的正向模式或反向 而之方向可從-基準圖像索引判定。然 模式時1不可能確認由二==:=:則:莫式或直接 像的實際方向。 1固基丰圖像索引指標的基準圖 ίί d ϊγ吏用多重基準圖像與-B圖像作為基準 方;ίϊ碼器不能單從基準圖像索引估計 万内貝戒,因為,即使它可以從 訊,確認-預設的正向/反向^丨的再映射貝 向索引順序之間的關係,但它、確::相二的正向’反 Β圖像的正向基準圖像與反向基^確目心解碼之 本發明中建議,在其淮闰你土 +圖像數罝。 指示其顯示順序:f引所指標的基準圖像中包括 序係以-圖像二 =二’, 每一基準圖像索:值表不的。因此,藉由比較 碼之B圖像的顯干値的母一基準圖像的顯示順序與欲編 向。 ,、^順序’可以輕易確認每-運動向量的方 量PMV二中^ 2 '則換式或直接模式時,計算預測運動向 若一欲編石馬區綠 時,該一鄰區鄰區塊為雙預測模式或直接模式 中,方向與沪^曰有二個運動向量。在此二個運動向量 必須用^中i ΪI1區塊之預測運動向量p M V方向相同者, 逆鼻。以下將詳細說明本發明建議的高效率Case No. 92100445 V. Description of the invention (8) Motion vector. If: ΐ With: The forward mode or reverse direction of a motion vector can be determined from the -reference image index. However, in mode 1 it is impossible to confirm the actual direction of two ==: =: then: Mo mode or direct image. 1 The benchmark map of Gujifeng's image index indicator uses multiple reference images and -B images as the reference side; the encoder cannot estimate Wannei ring from the reference image index alone, because even if it can From the news, confirm the relationship between the preset forward / reverse ^ re-mapping index order, but it: It is suggested in the present invention that the reverse-based decoding is accurate, and that you + the number of images in it. Indicate its display order: The reference images indexed by f include the sequence of -image two = two ', and each reference image is: the value is not shown. Therefore, by comparing the display order and the desired orientation of the mother-reference image of the B image of the comparison code. ", ^ Order 'can easily confirm the PMV of each-movement vector PMV 2 ^ 2' Then when the conversion or direct mode is used, the predicted motion is calculated. In the bi-prediction mode or the direct mode, there are two motion vectors in the direction and the center. In these two motion vectors, the predicted motion vector p M V in the block i ΪI1 must be used in the same direction. The high efficiency proposed by the present invention will be explained in detail below.

第12頁 I幽 1224930 _案號92100445_年月日 修正 五、發明說明(9) 預測運動向量PMV計算方法。為了說明方便,假設欲編碼 區塊為E ’鄰區塊為A、B、C、D,每一鄰區塊之二個運動 向量為MV1與MV2,如圖1所示。 2.1若鄰區塊的二個運動向量方向不同時 選取方向與欲編碼區塊預測運動向量pMV方向相同的運動 向量,然後經由中值運算計算預測運動向量PMV。 換言之,計算與區塊E正向預測運動向量相關的pjjV時,選 取一正向運動向量;而於計算與區塊E反向預測運動向量 相關的PMV時,則選取一反向運動向量。然後,經由中值 運算獲得每一方向之預測運動向量ρ Μ V。 2 · 2若鄰區塊的二個運動向量方向相同,且與預測運動向 量PMV之方向相同時 首先判 像,是 若二個 中之一 具有相 中值運 有相同 量之轉 像,可 若運動 圖像, 測運動 否為欲編碼區塊E所表示的基準圖像。 運動向iMVl、MV2都表示同一基準圖像,則選取其 (例如欲較先解碼之運動向量,或與預 動向量 算賴 同模式[MVi模式MV2模式]的運動向量),炎包括在 算中’用以計算預測運動向量pMv。,該/具 模式之運動向量,表示其轉動向 錄相同。此外’運動向細、準圖 能與欲編碼區塊Ε所表示的基準圖像相下、/相異。 向量Μ:1,中只有-個表示區塊基準 則將該-運動向量包括在中Page 12 Iyou 1224930 _ Case No. 92100445_ Year Month Day Amendment V. Description of the invention (9) Method for calculating PMV of predicted motion vector. For the convenience of explanation, it is assumed that the block to be coded is E ′ and the neighboring blocks are A, B, C, and D, and the two motion vectors of each neighboring block are MV1 and MV2, as shown in FIG. 1. 2.1 If the two motion vectors of neighboring blocks have different directions, select a motion vector with the same direction as the predicted motion vector pMV of the block to be coded, and then calculate the predicted motion vector PMV through the median operation. In other words, when calculating pjjV related to the forward prediction motion vector of block E, a forward motion vector is selected; and when calculating PMV related to the backward prediction motion vector of block E, a reverse motion vector is selected. Then, the predicted motion vector ρ MV in each direction is obtained through a median operation. 2 · 2 If the two motion vectors of the neighboring block have the same direction and the same direction as the predicted motion vector PMV, the image will be judged first. If one of the two has a phase median and the same amount of rotation, A moving image is used to measure whether the motion is a reference image represented by the block E to be coded. The motion directions iMVl and MV2 both represent the same reference image, then select it (for example, the motion vector to be decoded first, or the motion vector in the same mode as the pre-motion vector [MVi mode MV2 mode]), and inflammation is included in the calculation. 'To calculate the predicted motion vector pMv. The motion vector of this mode has the same rotation direction. In addition, the motion direction is fine and the quasi-picture can be different from / different from the reference picture represented by the block E to be coded. Vector M: 1, only one of them represents the block reference, then this motion vector is included in

向量PMV。 <开〒,用以J im 第13頁 1224930Vector PMV. < Open for J im page 13 1224930

將鄰區塊之二運動向量Mn 表示的基準圖像異於區塊E 為0 ;假定該鄰區塊 PMV。 ^ 侍欲編碼區塊之預測運動向量 r二π顯量時,則從關聯的基 是否與預測運動方運=量之方向’ *決定它們 近-e的m。如此;使=!=,错此而可獲得較接 (.m)範圍縮小,繼運動向量差_ 另一方面i若鄰區塊有一個運動向量時,則從一關聯基 圖像之顯示順序獲得該運動向量之方向資訊。若該運動向 量之方向異於預測運動向量之方向,則將該運動向量設^ 為0 ’並假設該一鄰區塊表示的基準圖像異於欲編碼區塊 所表示的基準圖像,並經由包括零運動之中值運算,獲得 預測運動向量。 ^ 此外,若鄰區塊有一個運動向量時,則從一關聯基準圖像The reference image represented by the motion vector Mn of the neighboring block is different from block E as 0; it is assumed that the neighboring block PMV. ^ When the predictive motion vector r of the waiter-coded block is significant, then it depends on whether the associated base is related to the predictive motion side = the direction of the amount '* to determine them near -e m. So; make =! =, Wrong to get a narrower (.m) range, following the motion vector difference_ On the other hand, if there is a motion vector in the neighboring block, the display order from an associated base image Get the direction information of the motion vector. If the direction of the motion vector is different from the direction of the predicted motion vector, set the motion vector to 0 ′ and assume that the reference image represented by the adjacent block is different from the reference image represented by the block to be encoded, and By including a zero-motion median operation, a predicted motion vector is obtained. ^ In addition, if a neighboring block has a motion vector,

第14頁 1224930 _案號92100445_年月日__ 五、發明說明(11) 之顯示順序獲得該運動向量之方向資訊。若該運動向量之 方向與預測運動向量之方向相同,則將該運動向量包括在 中值運算中,以計算預測運動向量。 從以上說明可知,本發明提供一種對基準圖像指定方向資 訊的方法,與一種判定基準圖像方向的方法,其中對每一 基準圖像指定可幫助獲得運動向量方向資訊的顯示順序資 訊。因此,可以獲得從目前欲編碼區塊至每一基準圖像的 方向資訊。 此外,本發明提供一種預測運動向量計算方法,其中,欲 編碼區塊周圍的某一鄰區塊因為一 B圖像之雙預測模式或 直接模式而有二個運動向量時,則定義欲在中值運算中使 用的運動向量。因此,可使用鄰區塊之運動向量資訊預測 欲編碼區塊之預測運動向量PMV,並可改進編碼效率。 以上雖然揭示本發明之較佳實施例作為舉例之用,但熟悉 此類技術之人士應可理解,其中可作多種修改與增減而不 脫離本發明所附申請專利範圍中揭示的範圍與精神。Page 14 1224930 _Case No. 92100445_Year Month Day__ V. Display order of invention description (11) Obtain the direction information of the motion vector. If the direction of the motion vector is the same as the direction of the predicted motion vector, the motion vector is included in the median operation to calculate the predicted motion vector. As can be seen from the above description, the present invention provides a method for specifying direction information on a reference image and a method for determining the direction of a reference image, wherein each reference image is specified with display order information that can help obtain direction information of motion vectors. Therefore, the direction information from the current block to be coded to each reference image can be obtained. In addition, the present invention provides a method for calculating a predicted motion vector. When a neighboring block around a block to be coded has two motion vectors due to a B-picture bi-prediction mode or a direct mode, it is defined that The motion vector used in the value operation. Therefore, the motion vector information of neighboring blocks can be used to predict the predicted motion vector PMV of the block to be coded, and the coding efficiency can be improved. Although the above describes the preferred embodiment of the present invention as an example, those skilled in the art should understand that various modifications and additions can be made without departing from the scope and spirit disclosed in the scope of the patent attached to the present invention. .

第15頁 1224930 _案號92100445_年月曰 修正_ 圖式簡單說明 [圖式簡單說明] 圖1顯示使用鄰區塊A、B、C之運動向量計算區塊E之預測 運動向量。 [元件符號對應表] · A鄰區塊 _ B鄰區塊 C鄰區塊 D鄰區塊 E欲編碼區塊Page 15 1224930 _Case No. 92100445_ Year Month Revision _ Brief Description [Schematic Description] Figure 1 shows the prediction motion vector of block E using the motion vectors of neighboring blocks A, B, and C. [Element symbol correspondence table] · A neighbor block _ B neighbor block C neighbor block D neighbor block E block to be coded

第16頁Page 16

Claims (1)

1224930 ^9210Π44Β /、、申睛專利範圍 方種對基準圖像指定方向資訊 方法,以提供由基準圖像索弓_為5亥基準圖像特徵之 訊,复由分本 ♦像京引和標的基準圖傻的太而次 ^ m 4 tf m ^ ^ , # 定“^分別,之基準圖像決 示順序資^ 〜y獲知每一基準圖像之顯 堍夕肋比較獲侍的顯示順序資訊與一目前铲绝踩π 鬼之顯示順序資訊;並以比較結」人、、扁馬區 圖像相對該目前欲編碼區塊之方:為基礎,決定每-基準 3·如申請專利範圍第3項所述之方法,其中 之顯示順序資訊係取自一圖像順序計數(p〇c)值。土,回象 4· 一種針對欲編碼區塊計算其預測運動向量(pMv)之方 法’其中係使用各鄰區塊之運動向量執行一中值運算,此 方法之步驟包括: a) 若該等鄰區塊各有多個運動向量,則先獲得該等鄰區塊 運動向量指向的基準圖像的方向資訊;以及 b) 參照所獲得的方向資訊,選取各該鄰區塊之運動向量之 一,並執行包括所選取運動向量之中值運算,以獲得欲編 碼區塊之預測運動向量。 5.如申請專利範圍第5項所述之方法,其中該步鱗a)進 包括以下步驟:比較各該鄰區塊運動向量指向的基準,$ 的顯示順序資訊與欲編碼區塊的顯示順序資訊,以決定^ 1224930 _案號92100445_年月日__ 六、申請專利範圍 等運動向量之方向資訊。 6. 如申請專利範圍第5項所述之方法,其中,若該等鄰區 塊之一有二個方向相異之運動向量時,該步驟b)進而包括 以下步驟:從二個運動向量中,選取方向與該預測運動向 量方向相同者,並執行包括所選取運動向量之中值運算, 以獲得該預測運動向量。 7. 如申請專利範圍第5項所述之方法,其中,若該等鄰區 塊之一有二個運動向量,其方向相同但與該預測運動向量 方向相異之時,該步驟b)進而包括以下步驟:將該二運動 向量設定為0 ;假定該一鄰區塊表示的基準圖像異於欲編 碼區塊所表示的基準圖像;並執行包括該零運動之中值運 算,以獲得該預測運動向量。 8. 如申請專利範圍第5項所述之方法,其中,若該等鄰區 塊之一有二個運動向量Μ VI與M V2,其方向相同並與該預測 運動向量方向相同,且該二運動向量Μ VI與MV2表示同一基 準圖像時,該步驟b)進而包括以下步驟b-Ι ):選取該二運 動向量MV1、MV2之一,並執行包括所選取運動向量之中值 運算,以獲得該預測運動向量。 9 ·如申請專利範圍第9項所述之方法,其中該步驟b - 1 )進 而包括步驟b-2):從該二運動向量MV1、MV2中,選取一個 欲較先解碼者,並執行包括所選取運動向量之中值運算, 以獲得該預測運動向量。 10.如申請專利範圍第9項所述之方法,其中該步驟b-1) 進而包括步驟b-2):從該二運動向量MV 1、MV2中,選取一1224930 ^ 9210Π44B /, Shen Jing patent range methods to specify direction information on the reference image to provide information from the reference image cable _ for the feature of the 5 Hai reference image. The reference picture is too silly ^ m 4 tf m ^ ^, # # "Respectively, the reference image decision sequence information ^ ~ y learns the display order information of each reference image compared to the display order information Based on the display sequence information of the current demons that have not stepped on π ghosts; and based on the comparison of the images of people, flat horses, and the current block to be coded: as a basis, determine the per-baseline 3. The method according to item 3, wherein the display order information is obtained from an image order count (poc) value. Figure 4: A method for calculating the predicted motion vector (pMv) of a block to be coded, where a median operation is performed using the motion vectors of neighboring blocks. The steps of this method include: a) if The neighboring blocks each have multiple motion vectors, then the direction information of the reference image pointed by the neighboring block motion vectors is obtained first; and b) referring to the obtained direction information, selecting one of the motion vectors of each neighboring block And perform a median operation including the selected motion vector to obtain a predicted motion vector for the block to be encoded. 5. The method according to item 5 of the scope of patent application, wherein step a) further includes the following steps: comparing the reference points of the motion vectors of the neighboring blocks, the display order information of $ and the display order of the block to be encoded Information to determine ^ 1224930 _ case number 92100445 _ year month day __ Six, direction information of motion vectors such as patent application scope. 6. The method according to item 5 of the scope of patent application, wherein if one of the neighboring blocks has two motion vectors with different directions, step b) further includes the following steps: from the two motion vectors , Selecting a direction that is the same as the direction of the predicted motion vector, and performing a median operation including the selected motion vector to obtain the predicted motion vector. 7. The method according to item 5 of the scope of patent application, wherein if one of the neighboring blocks has two motion vectors with the same direction but different directions from the predicted motion vector, step b) and then The method includes the following steps: setting the two motion vectors to 0; assuming that the reference image represented by the adjacent block is different from the reference image represented by the block to be coded; and performing a median operation including the zero motion to obtain The predicted motion vector. 8. The method according to item 5 of the scope of patent application, wherein if one of the neighboring blocks has two motion vectors M VI and M V2, their directions are the same and the direction of the predicted motion vector is the same, and the two When the motion vector MV VI and MV2 represent the same reference image, this step b) further includes the following steps b-1): selecting one of the two motion vectors MV1 and MV2, and performing an operation including the median value of the selected motion vector to This predicted motion vector is obtained. 9 · The method as described in item 9 of the scope of patent application, wherein step b-1) further includes step b-2): from the two motion vectors MV1 and MV2, select one to be decoded first and execute the include The median of the selected motion vector is calculated to obtain the predicted motion vector. 10. The method according to item 9 of the scope of patent application, wherein step b-1) further comprises step b-2): from the two motion vectors MV 1, MV2, select one 第18頁 1224930Page 18 1224930 - 條正____ 2模式UV1模式或MV2模式)與該預測運動向量之模式相同 者,並執行包括所選取運動向量之中值 測運動向量。 -r k 1 ·如申哨專利範圍第5項所述之方法,其中,若該等鄰 區塊之一f二個運動向量仏^與評?,其方向相同並與該預 測運動向I方向相同,但該二運動向量MV1與MV2中,僅有 一個表示欲編碼區塊所表示的基準圖像時,該步驟幻進而 包括以下步驟:從該二運動向量MV1、MV2中,選取一個表 示欲編碼區塊所表示基準圖像的運動向量,並執行包括所 選取運動向量之中值運算,以獲得該預測運動向量。 12·如申請專利範圍第5項所述之方法,其中,若該等鄰 區塊之一有二個運動向量MV1與MV2,其方向相同並與該預 測運動向量方向相同,但該二運動向量MV1與MV2都非表示 欲編碼區塊所表示的基準圖像,且分別表示不同的基準圖 像時,該步驟b )進而包括以下步驟:從該二運動向量 回 MV1、MV2中,選取一個表示基準圖像最接近欲編碼區塊所 表示基準圖像的運動向量,並執行包括所選取運動向量之 中值運算,以獲得該預測運動向量。 13·如申請專利範園第5項所述之方法,其中,若該等鄰 區塊之一有二個運動向量MV1與MV2,其方向相同並與讀 測運動向量方向相同,但該二運動向量MV1與MV2都非Z預 欲編碼區塊所表示的基準圖像,且分別表示不同的基準^ 像時’該步驟b )進而包括以下步驟:從該二運動向量 MV1、MV2中,選取〆個表示基準圖像最接近目前欲編瑪圖-Barbara ____ 2 mode (UV1 mode or MV2 mode) is the same as the mode of the predicted motion vector, and the motion vector including the median of the selected motion vector is executed. -r k 1 · The method as described in item 5 of the patent claim range, wherein if one of the neighboring blocks f two motion vectors 仏 ^ and evaluation? , Its direction is the same and the same as the predicted motion direction I, but when only one of the two motion vectors MV1 and MV2 represents the reference image represented by the block to be encoded, this step further includes the following steps: Among the two motion vectors MV1 and MV2, a motion vector representing a reference image represented by a block to be coded is selected, and a median operation including the selected motion vector is performed to obtain the predicted motion vector. 12. The method according to item 5 of the scope of patent application, wherein if one of the neighboring blocks has two motion vectors MV1 and MV2, the directions are the same and the direction of the predicted motion vector is the same, but the two motion vectors When neither MV1 nor MV2 represents the reference image represented by the block to be coded, and each represents a different reference image, step b) further includes the following steps: from the two motion vectors back to MV1 and MV2, select one representation The reference image is closest to the motion vector of the reference image represented by the block to be encoded, and performs a median operation including the selected motion vector to obtain the predicted motion vector. 13. The method as described in item 5 of the patent application park, wherein if one of the neighboring blocks has two motion vectors MV1 and MV2, the directions are the same and the direction of the read motion vector is the same, but the two motions Both the vectors MV1 and MV2 are not the reference images represented by the Z pre-encoded blocks, and each represents a different reference image. This step b) further includes the following steps: From the two motion vectors MV1 and MV2, select 〆 Indicates that the reference image is closest to the current map 1224930 _案號92100445_年月日__ 六、申請專利範圍 像的運動向量,並執行包括所選取運動向量之中值運算, 以獲得該預測運動向量。 14. 如申請專利範圍第5項所述之方法,其中,若該等鄰 區塊之一有一個運動向量,其方向與該預測運動向量方向 相異時,該步驟b )進而包括以下步驟:將該一鄰區塊之該 一運動向量設定為0 ;假定該一鄰區塊表示的基準圖像異 於欲編碼區塊所表示的基準圖像;並執行包括該一鄰區塊 之零運動的中值運算,以獲得該預測運動向量。 15. 如申請專利範圍第5項所述之方法,其中,若該等鄰 區塊之一有一個運動向量,其方向與該預測運動向量方向 相同時,該步驟b )包括以下步驟:執行包括該一鄰區塊之 運動向量的中值運算,以獲得該預測運動向量。1224930 _Case No. 92100445_ 年月 日 __ VI. Patent Application Scope Motion vectors and perform median operations including the selected motion vector to obtain the predicted motion vector. 14. The method according to item 5 of the scope of patent application, wherein if one of the neighboring blocks has a motion vector whose direction is different from the direction of the predicted motion vector, step b) further includes the following steps: Set the motion vector of a neighboring block to 0; assume that the reference image represented by the neighboring block is different from the reference image represented by the block to be coded; and perform zero motion including the neighboring block To obtain the predicted motion vector. 15. The method according to item 5 of the scope of patent application, wherein if one of the neighboring blocks has a motion vector whose direction is the same as the direction of the predicted motion vector, step b) includes the following steps: The median operation of the motion vector of the adjacent block to obtain the predicted motion vector. 第20頁Page 20
TW92100445A 2002-07-18 2003-01-10 Calculation method for prediction motion vector TWI224930B (en)

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
KR20020042204 2002-07-18

Publications (2)

Publication Number Publication Date
TW200402238A TW200402238A (en) 2004-02-01
TWI224930B true TWI224930B (en) 2004-12-01

Family

ID=34567616

Family Applications (1)

Application Number Title Priority Date Filing Date
TW92100445A TWI224930B (en) 2002-07-18 2003-01-10 Calculation method for prediction motion vector

Country Status (2)

Country Link
CN (1) CN100512444C (en)
TW (1) TWI224930B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9124898B2 (en) * 2010-07-12 2015-09-01 Mediatek Inc. Method and apparatus of temporal motion vector prediction
RU2626013C1 (en) 2011-10-19 2017-07-21 Кт Корпорейшен Video signal decoding method
CN103338372A (en) * 2013-06-15 2013-10-02 浙江大学 Method and device for processing video

Also Published As

Publication number Publication date
TW200402238A (en) 2004-02-01
CN1856107A (en) 2006-11-01
CN100512444C (en) 2009-07-08

Similar Documents

Publication Publication Date Title
TWI690200B (en) Intra merge prediction
JP4331776B2 (en) Predicted motion vector calculation method and motion vector prediction method
JP4625132B2 (en) Derivation method of direct mode motion vector
CN103238319B (en) Derivation motion vector predictor or the method and device of motion vector predictor candidate item
JP2019041418A5 (en)
WO2016165617A1 (en) Method and apparatus for deriving temporal motion vector prediction
JP2018511997A (en) Image prediction method and related apparatus
JP2008172833A (en) Method of predicting block
TW201136325A (en) Selectively motion vector prediction method, motion estimation method and device thereof applied to scalable video coding system
JP7223111B2 (en) Motion vector prediction method and apparatus, encoder and decoder
JP5858193B2 (en) Image decoding apparatus, image decoding method, and image decoding program, and receiving apparatus, receiving method, and receiving program
TWI224930B (en) Calculation method for prediction motion vector
JP5679024B2 (en) Image decoding apparatus, image decoding method, and image decoding program, and receiving apparatus, receiving method, and receiving program
CN110677673A (en) Interframe coding method, interframe decoding method, equipment and storage medium thereof

Legal Events

Date Code Title Description
MM4A Annulment or lapse of patent due to non-payment of fees