1283133 •九、發明說明： 、【發明所屬之技術領域】 本發明係有關於一種移動估計方法，更詳而言之，係 有關於一種於針對視訊物件平面中形狀資訊予以編碼的移 動估計方法。 【先前技術】 MPEG(Moving Picture Expert Group ;動態圖形專家小 ^ 組）是專責制定有關移動壓縮編碼標準的工作組，所制定白勺 標準是國際通用標準，亦即，MPEG標準，該標準係由視 頻、音頻以及系統三部份組成。其中，MPEG-1是影音光 碟（Video Compact Disc ; VCD)的視頻圖像壓縮標準； MPEG_2 是數位視訊影碟（Digital Video Disc ; DVD)/超級 影音光碟的視頻圖像壓縮標準；而MPEG-4則是網路視頻 圖像壓縮標準之一，具有壓縮比高、成像清晰等特徵，與 MPEG-1及MPEG-2相比，MPEG-4更加注重多媒體系統 ~·的交互性和靈活性。 - 再者，MPEG-4係為第一個以物件為基礎 (Object-Based)的壓縮方法。根據MPEG-4標準’每一個書 面都是由視訊物件平面（Video Object Plane ; VOP)所構 成，可以單獨或聯合方式對場景（scene)中的視訊物件（AV Object)進行編碼、存儲以及傳輸操作。視訊物件在各視訊 物件平面中係由形狀（shape)、紋理（texture)以及移動 (motion)等資訊組成，MPEG-4標準中的每個視訊物件平面 係分為形狀（Shape)資訊以及紋理（Texture)資訊以分別予 5 18769(修正本) 1283133 '，儲存，請參閱第m⑺圖，該第以及則係分別 -顯示了一視訊物件平面相應之形狀資訊u晝面及紋理資 訊上2晝面’其中，該視訊物件平面對應之形狀資訊⑴系 由複數形狀區塊（Binary AlphaBlock ; BAB)11〇所構成， 相對地’該視訊物件平面對應之紋理資訊12則由複數紋理 巨集塊（M_Block)120所構成，且於各紋理巨集塊12〇 中係包括有4個紋理區塊。再者，該形狀資訊n中各形狀 -區塊11G係與該紋理資訊12中各紋理巨集塊12()的位置 _互對應。 清配合苓閱第2圖，若針對該欲編碼視訊物件平面j 之形狀資訊11進行編碼，則需對該形狀資訊n中的每一 個形狀區塊110逐個進行編碼，具體而言，係於該欲編碼 視訊物件平面1中定義一欲編碼區塊U1，並於相對該欲 編碼視訊物件平面1之前一時刻視訊物件平面，亦即，參 考視訊物件平面2中尋找與該欲編碼區塊lu最接近的形 -籲狀區塊以作為參考區塊2U，這個搜索過程即稱為形狀移 .動估計。 ^ MPEG_4編碼的數據壓縮效率依賴於移動估計的準確 性，而移動估計的準確性則依賴於所需的運算時間。請配 合參閱第3A圖，習知的移動估計方法係首先設定欲編碼 區塊111之候選向量區塊，亦即，鄰區塊i i 3、鄰區塊115、 鄰區塊117、鄰區塊丨23、鄰區塊125以及鄰區塊127，其 中，該鄰區塊113、鄰區塊115及鄰區塊117係為該欲編 碼區塊ill所對應候選向量形狀區塊（16*16像素），且將該 18769(修正本) 6 1283133 -·鄰區塊113、鄰區塊115及鄰區塊117之移動向量分別定 '叔為（MVsl) (MVs2)以及（MVs3);而鄰區塊123、鄰區塊 125及鄰區塊127則為該欲編碼區塊111所對應之候選向 量紋理區塊（8*8像素），且將該鄰區塊123、鄰區塊'125及 鄰區塊127之移動向量分別定義為（MV1)、（mv2)以及 (MV3) ’於執行移動估計作#時，係首先從該候選向量列 表{ MVsl ’ MVs2 ’ MVs3，MV1，MV2，MV3 }中選取第 ，一個有效之移動向量，俾作為該欲編碼區塊lu之形狀移 .•動向量預測值（Motion Vect〇r Predict〇r f〇r ，以下簡 稱MVPs) ’即如第3 (B)圖所示之移動向量（mvs4以 及MV4 ) ’爾後，計算該Mvps對應之參考區塊2}!與該 欲編碼區塊111間的絕對差值總和（Sum 〇f Absolve1283133 • Nine, invention description: [Technical field to which the invention pertains] The present invention relates to a motion estimation method, and more particularly to a motion estimation method for encoding shape information in a plane of a video object. [Prior Art] MPEG (Moving Picture Expert Group) is a working group dedicated to the development of mobile compression coding standards. The standard is the international standard, that is, the MPEG standard. Video, audio and system are three parts. Among them, MPEG-1 is a video image compression standard for Video Compact Disc (VCD); MPEG_2 is a video image compression standard for digital video disc (Digital Video Disc; DVD)/Super Video CD; and MPEG-4 It is one of the network video image compression standards, featuring high compression ratio and clear imaging. Compared with MPEG-1 and MPEG-2, MPEG-4 pays more attention to the interactivity and flexibility of multimedia systems. - Furthermore, MPEG-4 is the first Object-Based compression method. According to the MPEG-4 standard, each written by a Video Object Plane (VOP) can encode, store, and transfer video objects in a scene, either individually or in combination. . The video object is composed of information such as shape, texture and motion in the plane of each video object. Each video object plane in the MPEG-4 standard is divided into shape information and texture ( Texture) Information is given to 5 18769 (Revised) 1283133 ', for storage, please refer to the m(7) diagram, which is respectively - displays the corresponding shape information of the video object plane, and the texture information. ' The shape information corresponding to the plane of the video object (1) is composed of a plurality of shape blocks (Binary AlphaBlock; BAB) 11 相对, and the texture information 12 corresponding to the plane of the video object is composed of a plurality of texture macro blocks (M_Block). 120 is constructed, and four texture blocks are included in each texture macroblock 12A. Furthermore, each shape-block 11G in the shape information n corresponds to the position_ of each texture macroblock 12() in the texture information 12. Referring to FIG. 2, if the shape information 11 of the video object plane j to be encoded is encoded, each shape block 110 in the shape information n needs to be encoded one by one, specifically, To define a video block U1 defined in the plane 1 of the video object, and to view the video object plane at a moment before the plane 1 of the video object to be encoded, that is, the reference video object plane 2 is the most searched for the block to be coded. The close shape-call block is used as the reference block 2U. This search process is called shape shifting and motion estimation. ^ The data compression efficiency of MPEG_4 encoding depends on the accuracy of the motion estimation, and the accuracy of the motion estimation depends on the required computation time. Referring to FIG. 3A, the conventional motion estimation method first sets a candidate vector block to be coded block 111, that is, a neighboring block ii 3, a neighboring block 115, a neighboring block 117, and a neighboring block. 23. The neighboring block 125 and the neighboring block 127, wherein the neighboring block 113, the neighboring block 115, and the neighboring block 117 are the candidate vector shape blocks corresponding to the block to be coded ill (16*16 pixels). And the motion vector of the 18769 (corrected) 6 1283133 - neighboring block 113, the neighboring block 115, and the neighboring block 117 are respectively determined as 'unsired (MVsl) (MVs2) and (MVs3); and the neighboring block 123, the neighboring block 125 and the neighboring block 127 are candidate vector texture blocks (8*8 pixels) corresponding to the block to be encoded 111, and the neighboring block 123, the neighboring block '125 and the neighboring area The motion vectors of block 127 are defined as (MV1), (mv2), and (MV3)', respectively, when performing motion estimation for #, first selecting from the candidate vector list { MVsl ' MVs2 ' MVs3, MV1, MV2, MV3 } First, a valid motion vector, 俾 as the shape shift of the block to be coded. • Motion vector predictor (Motion Vect〇r Predict〇rf〇r Hereinafter, MVPs) 'is the motion vector (mvs4 and MV4) shown in the third (B) diagram, and then calculates the sum of the absolute differences between the reference block 2 corresponding to the Mvps and the block 111 to be coded. (Sum 〇f Absolve
Difference，以下簡稱SAD)，並判斷SAD是否小於匹配誤 羞閾值（業界規定之標準值），若該sad小於匹配誤差閾 2則將MVPs作為§亥欲編碼區塊j i j最終的形狀移動向 量（M〇t_ Vector 〇f Shape，以下簡稱 MVs);而當 sad 大 於匹配誤差閾值，則於—預設範圍内，亦即，以·s為 中心’ ±16像素為半徑的範圍㈣行搜索，俾得到最終的 然’上述技術的缺點係在於MVPs的確定係從候選列 表{MVs卜MVs2，MVs3，_，體，则）中選取第一 個有效的移動向量，該種估計方法具有較大的隨機性，益 法保證所選取的第-個有效移動向量是否恰#，如此將大 大增加系統運算量，亦會提高硬體設備的複雜度；此外， 18769(修正本) 7 1283133 採用上述估計方法所計算得出的MVPs誤差較大，精度性 較低；再者，若依據該MPVs對應計算出之SAD係大=匹 配誤差閾值時，因形狀移動向量預測值MpVs定位的精度 不局則必須於土 16像素的較大範圍内再次進行搜索，此 舉不僅增加了計算的複雜度，且耗費時間較多，進而影響 即時編碼作業之流暢性。 θ 綜上所述，如何開發一種形狀移動估計方法，能以最 小的運算複雜性達到最大的結果準輕，實為與視訊編碼 相關的領域者亟需研究之重要課題。 【發明内容】 鑒於上述習知技術之缺點，本發明主 一種可提高運算速度之視訊編碼之㈣估計方 曰本^月之另一目的係在於提供一種可提高形狀移動 向1預測值的精確度之視訊編碼之移動估計方法。 為達上述目的及其他相關目的，本發明即提供一種視 訊編碼之移動估計方法，本發明之視訊編碼之移動估計方 係用以針對欲編碼之視訊物件平面⑺de。〇bject _ ’ V〇P) ’計算其形狀移動向量(Motion Vector of #+ bT »MVS) ’ 4方法包括以下步驟··⑴設定欲編碼區塊 占二之第-候選向量區塊組與第二候選向量區塊組；⑺ 偵測對應該欲編石馬區# 楚 有效之移動向量，若否向量區塊組^否存在 驟右否則進至步驟（3)，若是，則進至步 Η 、測對應該欲編碼區塊之第二候選向量區塊組中 有放之移動向量’若否，即設定該欲編碼區塊之 18769(修正本) 8 1283133 -形狀移動向量預測值（Motion vector Predictor for shape; ' MVPs)為〇，並進至步驟（5)，若是，則進至步驟（句 分析所偵測到的有效之移動向量，並依據所分析出的有效 之私動向i個數，设置該欲編碼區塊之形狀移動向量預測 值，（5)計算該形狀移動向量預測值所對應之區塊與該欲編 碼區塊間的絕對差值總和（Sum 〇f Abs〇lute Difference ; SAD)，並判斷該絕對差值總和是否小於匹配誤差標準閾 值，若是，即得出該欲編碼區塊之形狀移動向量（M〇ti〇n — •Vector of Shape ;MVs)等於該形狀移動向量預測值；若否， 則進至步驟（6);以及（6)於一預設範圍内搜索與該欲編碼區 塊間的絕對差值總和（s A D)為最小的形狀移動向量預測 值’藉而得出該欲編碼區塊之形狀移動向量。 中，5亥欲編碼區塊係為該視訊物件平面中一形狀區 塊（Binary Alpha B1〇ck ; BAB)。再者，該第一候選向量區Difference, hereinafter referred to as SAD), and determine whether the SAD is less than the matching false shading threshold (standard value specified by the industry). If the sad is less than the matching error threshold 2, the MVPs are used as the final shape motion vector of the jihai coding block jij. 〇t_ Vector 〇f Shape, hereinafter referred to as MVs); and when sad is greater than the matching error threshold, then within the pre-set range, that is, the range of ±16 pixels as the radius of the range (4) is searched, Finally, the shortcoming of the above technique is that the determination of the MVPs selects the first valid motion vector from the candidate list {MVsb MVs2, MVs3, _, body, then), and the estimation method has greater randomness. , the benefit method ensures that the selected first effective motion vector is #, which will greatly increase the amount of system operation, and will also increase the complexity of the hardware device; in addition, 18769 (Revised) 7 1283133 is calculated by the above estimation method. The obtained MVPs have large errors and low precision. Moreover, if the SAD system large = matching error threshold is calculated according to the MPVs, the accuracy of the shape motion vector predictor MpVs is not in the way. It is necessary to search again in a large range of 16 pixels of soil, which not only increases the computational complexity, but also takes more time, thereby affecting the smoothness of the instant coding operation. θ In summary, how to develop a shape motion estimation method, which can achieve the maximum result with the smallest computational complexity, is an important subject that needs to be studied in the field related to video coding. SUMMARY OF THE INVENTION In view of the above-mentioned shortcomings of the prior art, the fourth aspect of the present invention is to improve the accuracy of the shape motion to the predicted value of the image. A method of motion estimation for video coding. To achieve the above and other related objects, the present invention provides a motion estimation method for video coding. The motion estimation method of the video coding of the present invention is used for the video object plane (7) to be encoded. 〇bject _ ' V〇P) 'Calculate the shape motion vector (Motion Vector of #+ bT »MVS) ' 4 methods include the following steps · (1) set the block to be coded to occupy the second - candidate vector block group and Two candidate vector block groups; (7) detecting the moving vector corresponding to the desired Shima area #楚, if no vector block group ^ there is a right, otherwise proceed to step (3), and if so, proceed to step And measuring the motion vector of the second candidate vector block group corresponding to the block to be coded, if not, setting the 18769 (corrected version) of the block to be coded 8 1283133 - shape motion vector predictor (Motion vector Predictor for shape; ' MVPs) is 〇, and proceeds to step (5), and if so, proceeds to step (the effective motion vector detected by the sentence analysis, and according to the analyzed effective private movement to i number, Setting a shape motion vector predictor of the block to be coded, and (5) calculating a sum of absolute differences between the block corresponding to the shape motion vector predictor and the block to be coded (Sum 〇f Abs〇lute Difference; SAD ) and determine whether the sum of the absolute differences is less than Matching the error standard threshold, if so, the shape motion vector (M〇ti〇n - • Vector of Shape; MVs) of the block to be coded is equal to the shape motion vector predictor; if not, proceed to step (6) And (6) searching for a shape movement vector predictor value of the sum of absolute differences (s AD) between the block to be encoded and a minimum within a predetermined range to obtain a shape movement of the block to be coded In the vector, the 5H code block is a shape block in the video object plane (Binary Alpha B1〇ck; BAB). Furthermore, the first candidate vector area
塊組係由複數相臨於該欲編碼區塊之形狀區塊所構成，而 該第二候選向量區塊組係由複數相臨於該欲編碼區塊之紋 理區塊所構成，且該形狀區塊的尺寸係為16*16像素，該 紋理區塊的尺寸係為8*8像素。再者，於該步驟（6)中係以 該計算出之形狀移動向量預測值（MVPs)為中心，且於預定 數量的像素（pixel)的區域内進行搜索。 較佳地，該第一候選向量區塊組係由三個相鄰於該欲 編碼區塊之形狀區塊（Binary Alpha Block ; BAB)所構成， 而該第二候選向量區塊組係由三個相鄰於該欲編碼區塊之 紋理區塊所構成。且於步驟（4)中復包括以下步驟··（4_υ W769(修正本） 1283133 '分析有效之移動向量的個數 .·卜則進至步驟（4斗若錢=之量的個數為 至步驟㈣，若有效 動向㈣個數為2，則進 (4-4) ； (4 ^ 砂勒向里的個數為 (4)，（4·2Μ置該欲編碼區塊 2 ㈣進至步驟 该有效之移動向量；（4_ :夕動向量預測值等於 置=等於該二個有效之移動向量的均值：;動向 以奴、4碼區塊之形狀移動向量預測 :及（4-4)設 移動向量的中間值及均值之其中一、者荨於该二個有效之 不僅視訊編碼之移動估計方* 形狀移動向量預測值無法確，且於該 圍搜索最終之形狀移動向量，:==:可以較小地範 之功效。 ㈣里㈣相提以統運算速度 【實施方式】 、以下係藉由特定的具體實例說明本發明之實施方 ·:’熟悉此技藝之人士可由本說明書所揭示之内容輕易地 目,、解本冑明之其他優點與功纟。本發明亦可藉由其他不同 的具體實例加以施行或應用，本說明書中的各項細節亦可 基於不同觀點與應用，在不丨孛離本發明之精神下進行各種 修傅與變更。 晴參閱第4圖，其係為本發明之視訊編碼之移動估計 方法之運作流程圖。在此亦請配合第1A圖、第1B圖、第 2圖、弟3 A圖以及第3B圖詳細說明本發明之視訊編碼之 考夕動估計方法的運作流程。於本實施例中，該移動估計方 10 18769(修正本) 1283133 .法係用以針對欲編碼之視訊物件平面（vide〇 -Plane ; VOP) ’計算其形狀移動向量（M〇ti〇n 〇fThe block group is composed of a plurality of shape blocks adjacent to the block to be coded, and the second candidate vector block group is composed of a plurality of texture blocks adjacent to the block to be coded, and the shape is The size of the block is 16*16 pixels, and the size of the texture block is 8*8 pixels. Furthermore, in the step (6), the calculated shape motion vector predictors (MVPs) are centered, and the search is performed in a predetermined number of pixels. Preferably, the first candidate vector block group is composed of three shape blocks (Binary Alpha Block; BAB) adjacent to the block to be coded, and the second candidate vector block group is composed of three A texture block adjacent to the block to be coded. And in step (4), the following steps are further included: (4_υ W769 (Revision) 1283133 'Analyze the number of valid moving vectors. · Bu into the step (4 buckets if the amount of the amount = to Step (4), if the number of effective movements (four) is 2, then enter (4-4); (4 ^ The number of the inward is (4), (4·2 set the block to be coded 2 (4) to the step The effective motion vector; (4_: the eigenvector predictor is equal to set = equal to the mean of the two valid motion vectors:; the motion is slave, the shape of the 4 code block moves the vector prediction: and (4-4) One of the median and mean values of the motion vector is not only valid for the motion estimation of the video coding * the shape motion vector predictor is not valid, and the final shape motion vector is searched for the perimeter: :==: (4) In the following, the implementation of the present invention is explained by specific specific examples: [The person familiar with the art may disclose the contents of the present specification. Easily understand the other advantages and merits of Benming Ming. The various details and embodiments of the present invention can be implemented or applied in various different embodiments, and various modifications and changes can be made without departing from the spirit and scope of the invention. It is a flowchart of the operation of the motion estimation method for video coding of the present invention. Here, please also refer to FIG. 1A, FIG. 1B, FIG. 2, FIG. 3A and FIG. 3B for a detailed description of the video coding of the present invention. In the present embodiment, the motion estimation method 10 18769 (amendment) 1283133. The method is used to calculate the shape movement of the video object plane (vide〇-Plane; VOP) to be encoded. Vector (M〇ti〇n 〇f
Shape;以下簡稱為MVs)，其中，該形狀資訊^係由複數 -個尺寸為16*16像素（Pixei)之形狀區塊（BinaryAlpha Block ’ ΒΑΒ)11〇所組成，該紋理資訊12則由複數個尺寸 為16*16像素之紋理巨集塊（Macr〇B1〇ck)12〇所構成，且 於各紋理巨集塊12〇中復包括有4個尺寸為8*8像素之紋 理區塊。 _ 如圖所示，首先執行步驟S200,設定欲編碼區塊對應 之候選向量區塊，其中，該欲編碼區塊ln係為欲編碼視 σ孔物件平面1中一形狀區塊，且係將該欲編碼區塊1 1 1對 應之候選向量分為第一候選向量區塊組及第二候選向量區 塊組，且該第一候選向量區塊組係由複數相臨於該欲編碼 區塊111之形狀區塊所構成，於本實施例中，該第一候選 向量區塊組係包括分別位於該欲編碼區塊i i丨之左侧、頂 魯部以及右側頂部之鄰區塊113、鄰區塊i i 5以及鄰區塊 117，其可分別以MVsl、MVs2及MVs3標示；該第二候 選向量區塊組係由複數相臨於該欲編碼區塊U1之紋理區 塊所構成，請參閱第3A圖，其中，該欲編碼區塊1丨丨對 應之第一候選向量區塊組中所包含之各形狀區塊，亦即， 該鄰區塊113、鄰區塊Π5以及鄰區塊117所對應之紋理 巨集塊係分為該紋理巨集塊122、紋理巨集塊124以及紋 理巨集塊126，於本實施例中，該第二候選向量區塊組係 由上述各紋理巨集塊中對應的紋理區塊所構成，亦即，該 11 18769(修正本) 1283133 .第—候選向量區塊組係包括分別位於該欲編碼區塊111之 -左侧、頂部以及右側頂部之鄰區塊123、鄰區塊125以及 4區塊127所構成，其可分別以MV1、MV2及MV3標示 (即如第3A圖所示）。需說明者，本發明之於欲編碼區塊 111對應之候選向量區塊的個數及所處位置並不以此為 限，其中，該第一候選向量區塊組亦可分別由四個分別位 =該欲編碼區塊ln之左側、頂部、右側頂部以及左側頂 部之鄰區塊113、鄰區塊115、鄰區塊117以及鄰區塊ιι9 所組成，而該第二候選向量區塊組則可分別由四個分別位 ㈣欲編碼區塊m之左側、頂部、右側頂部以及左側頂 區塊123、鄰區塊125、鄰區塊127以及鄰區塊129 成（即如第3B圖所示），接著進至步驟S201。 ⑽=S;〇T判斷第—候選向量區塊組丨胸， 夂“ S3中疋否存在有效之移動向量，亦即，判斷 ^^該欲編碼區塊lu之形狀區塊中是 若否，則進至步物，若是，則進至步驟 於步驟S202中，於判斷出該第一候選 二^，MVs2，MVs3丨中並未存在有 ^旦且 接者判斷第二候選向量區塊組，夕動向置，則 否存在有效之移動向量，亦即 +，MV3丨中是 塊⑴之紋理區塊中是否存在有效目=於該欲編碼區 進至步驟謂，若是，則進至步驟咖動向量’若否，則 於步驟S203中，分析有 。 私動向量的個數，具體 12 、 18769(修正本） 1283133 而言，係分析第一候選向量區塊組{ MVsl，MVs2，MVs3 } 或第二候選向量區塊組{MV1，MV2，MV3}中存在之有 效的移動向量的個數，若有效之移動向量的個數為1，則 進至步驟S204 ;若有效之移動向量的個數為2，則進至步 驟S205;若有效之移動向量的個數為3,則進至步驟S206。Shape; hereinafter referred to as MVs), wherein the shape information ^ is composed of a plurality of 16*16 pixel (Pixei) shape blocks (BinaryAlpha Block ' ΒΑΒ) 11〇, and the texture information 12 is composed of plural numbers. The texture macroblock (Macr 〇 B1 〇 ck) 12 尺寸 is composed of 16*16 pixels, and four texture blocks of 8×8 pixels are included in each texture macroblock 12〇. As shown in the figure, step S200 is first executed to set a candidate vector block corresponding to the block to be coded, wherein the block to be coded is a shape block in the plane 1 of the object to be coded, and the system will The candidate vector corresponding to the block 1 1 1 is divided into a first candidate vector block group and a second candidate vector block group, and the first candidate vector block group is adjacent to the block to be coded by the complex number. The shape block of 111 is formed. In this embodiment, the first candidate vector block group includes adjacent blocks 113, adjacent to the left side, the top part, and the top part of the block to be coded, respectively. Block ii 5 and neighboring block 117, which may be denoted by MVs1, MVs2 and MVs3 respectively; the second candidate vector block group is composed of a plurality of texture blocks adjacent to the block to be coded U1, see FIG. 3A, wherein each shape block included in the first candidate vector block group corresponding to the block to be encoded, that is, the neighbor block 113, the neighbor block Π5, and the neighbor block 117 The corresponding texture macroblock is divided into the texture macroblock 122 and the texture macroblock 124. And the texture macroblock 126. In this embodiment, the second candidate vector block group is composed of corresponding texture blocks in each of the texture macroblocks, that is, the 11 18769 (corrected version) 1283133 . The first candidate vector block group includes adjacent blocks 123, neighboring blocks 125, and 4 blocks 127 located at the top left, top, and right sides of the block to be coded 111, respectively, which may be respectively MV1. MV2 and MV3 are marked (ie as shown in Figure 3A). It should be noted that the number and location of the candidate vector blocks corresponding to the block to be coded 111 in the present invention are not limited thereto, and the first candidate vector block group may also be respectively divided into four parts. Bit = the neighboring block 113, the neighboring block 115, the neighboring block 117, and the neighboring block ιι9 of the left side, the top, the top of the right side, and the top left side of the block to be encoded, and the second candidate vector block group Then, it can be respectively formed by four respective bits (4) to the left side, the top side, the top right side, and the left top block 123, the neighboring block 125, the neighboring block 127, and the neighboring block 129 of the block m (ie, as shown in FIG. 3B). Shown), then proceed to step S201. (10)=S; 〇T judges the first candidate vector block group, 夂 “S3 has a valid motion vector, that is, it is judged whether the shape block of the block to be coded is in the shape block. Then, the process proceeds to step S202, where it is determined that the first candidate two, MVs2, MVs3丨 does not exist, and the second candidate vector block group is determined. If there is a moving motion vector, there is no effective motion vector, that is, +, MV3丨 is whether there is a valid mesh in the texture block of block (1) = in the step of the code to go to the step, if yes, then go to the step If the quantity 'if no, then in step S203, the number of the private motion vectors is analyzed. Specifically, 12, 18769 (Revised) 1283133, the first candidate vector block group { MVsl, MVs2, MVs3 } or The number of valid motion vectors existing in the second candidate vector block group {MV1, MV2, MV3}, if the number of valid motion vectors is 1, then proceeds to step S204; if the number of valid motion vectors is If it is 2, the process goes to step S205; if the number of valid motion vectors is 3, then the process goes to step S206. .
於步驟S204中，設置MVPs等於該有效之移動向量， 例如若分析得出該第一候選向量區塊組{MVsl，MVs2， MVs3}中僅MVsl係為有效之移動向量，則令MVPs = MVsl，接著進至步驟S208。 於步驟S205中，設置MVPs等於該二個有效之移動 向量的均值，例如若分析得出該第一候選向量區塊組 { MVsl，MVs2，MVs3 }中MVsl及MVs2中係為有效之 移動向量，則令（MVPs)=((MVsl)+(MVs2))/2,接著進至 步驟S208。 於步驟S206中，設置MVPs等於該三個有效之移動 向量的中間值及均值之其中一者，例如若分析得出該第一 候選向量區塊組{ MVsl，MVs2，MVs3 }中 MVsl、MVs2 以及MVs3均為有效之移動向量，則取出第一候選向量區 塊組{ MVsl，MVs2，MVs3 }間的中間值MVs2作為MVPs， 或者對第一候選向量區塊組{ MVsl，MVs2，MVs3 }的 (MVsl)、（MVs2)以及（MVs3)進行計算，以將計算而求得 的均值（亦即，（(MVsl)+(MVs2)+(MVs3))/3 )作為（MVPs)， 接著進至步驟S208。 於步驟S207中，由於判斷出該第一候選向量區塊組 13 18769(修正本) 1283133 •以及第二候選向量區塊組均未存在有效之移動向量，則設 —置該欲編碼區塊in之形狀移動向量預測值（]^〇11〇11 .Vector Predictor f0r shape ;以下簡稱 Mvps；^ 〇，於本實 施例中，該（MVPs)係以二位元數值表示，亦即，（Mvp^ = (0,0)，並進至步驟S208。 於步驟S208中，於參考視訊物件平面2中找出Mvps 所對應之區塊，亦即參考區塊211，且計算該參考區塊2ΐι 與該欲編碼區塊111間的絕對差值總和（Sum 〇f Ab⑽1扯e • Difference;以下簡稱SAD),冑說明者，由於sad之運算 方法係為-般視訊編碼之技術，且亦非本案之技術特徵了 故在此將不予贅述’接著進至步驟S2〇9。 於步驟顯中，判斷SAD是否小於匹配誤差閑值， 於本實施例中’該匹配誤差閾值係為業界所規定的標準 值’即（16*AlphaTH) ’若否，則進行步驟_，若是，則 進行步驟S211。 於步驟議中，於預設之區域範圍内，亦即，以聽s =二並以例如加5個以及減5個像素為半徑的區域範 圍内搜索出與該欲編碼區塊lu間的SAD為最小的區塊， =料算公式以計算出該欲編碼區塊ιη最終的肠， ㈣中’該計算公式係為形狀移動向量差值= (MVs) 即可依據所取得的形狀移動向量差值進 = 的視訊編碼’並可結束本發明之視訊編碼之 矛夕動估計方法之處理步驟。 ’ 於步驟S211中’設置該欲編碼區塊m之形狀移動 14 18769(修正本） 等於體S，亦即，將該_作為該欲編碼區 現111取終的形狀移動向量，並 量進行視Μ石，日叮处“ 的形狀移動向 ::::= 之視訊編竭之移動估計 候選二=，本發明之移動估計方法係將習知技術中各 ^向1區塊按照類型整合成域之區塊組，並 塊組内存在之有效移動向量 · °° _b 要又以刊用不同的計瞀古 =ΓΓ碼區塊之形狀移動向量預測值，相較於習：技 欲編碼候f向量區塊中第一個有效之移動向量作為 人、、扁碼£塊的形狀料向量預測I，本發 量預測值的估算方式更為合理 1私動向 ，卜由於耗狀㈣向！預測值之精確度大合， 大而畲該形狀移動向量預測值不 门 較小的範圍來搜索該欲編碼區塊最終==丨可以 善系統之運作性能的目的。 《而達到改 ^上所述僅為本發明之較佳實施方式而已 限疋本發明之範圍，亦即，本發 卞用以 例如，本發明之韻郊魄成々 只上仍可做其他改變， 所、十： …’焉之移動估計方法之處理步驟S210 所述的搜索之預設範圍並不限 U>21〇 作為半徑，亦即，端視實施型能而:5個以及減5個像素 半徑。因此，舉凡熟習該項==同r像素個數作為 之精神與技術思想下所完成之一=離本發明所揭示 由後述之申請專利範圍所涵蓋。、效知飾或改變’仍應 18769(修正本) 15 1283133 【圖式簡單說明】 第1A及1B圖係為一實施例圖，复 訊編碼處理後對於一視訊物 ，、糸/刀別頻不習知視 資訊及紋理資訊； 〇p)所生成之之形狀 第2圖係用以說明視訊編 笼3Δ闰总田θ 私動估計原理示意圖； 弟3Α圖係用以顯示本發明之於 口 候選向量區塊之第-實施例示意圖；、…4及其相應之 第3Β圖係用以顯示本發明之於編碼區塊及其相 候逛向量區塊之第二實施例示意圖；以及 … 第4圖係為本發明之視訊編石^ ^ ^ ^ ^ ^ •圖。 k㈣估村法之運作 【主要元件符號說明】 11 110 :·111 113,115,117,119 ^ 12 120，122，124，126 123，125，127，129 2 211 S200〜S211 欲編碼視訊物件平面 形狀資訊（Shape) 形狀區塊（BAB) 預編碼區塊 鄰區塊（形狀區塊） 紋理資訊（Texture) 紋理巨集塊（Macro Block) 鄰區塊（紋理區塊） 參考視訊物件平面 蒼考區塊 步驟 16 18769(修正本)In step S204, the MVPs are set to be equal to the valid motion vector. For example, if only MVs1 in the first candidate vector block group {MVsl, MVs2, MVs3} is determined to be a valid motion vector, then MVPs = MVsl, Then it proceeds to step S208. In step S205, the MVPs are set to be equal to the mean of the two effective motion vectors. For example, if the motion vectors of the first candidate vector block group { MVsl, MVs2, MVs3 } are valid in the MVs1 and MVs2, Then, (MVPs) = ((MVsl) + (MVs2))/2, and then proceeds to step S208. In step S206, the MVPs are set to be equal to one of the intermediate values and the mean values of the three valid motion vectors, for example, if the MVs1, MVs2 in the first candidate vector block group { MVsl, MVs2, MVs3 } are analyzed, If MVs3 is a valid motion vector, the intermediate value MVs2 between the first candidate vector block group { MVsl, MVs2, MVs3 } is taken as MVPs, or for the first candidate vector block group { MVsl, MVs2, MVs3 } ( MVsl), (MVs2), and (MVs3) are calculated to calculate the mean (ie, ((MVsl) + (MVs2) + (MVs3))) / 3) as (MVPs), and then proceed to the step S208. In step S207, since it is determined that the first candidate vector block group 13 18769 (correction) 1283133 • and the second candidate vector block group have no valid motion vector, then the block to be coded is set to be in The shape motion vector predictor ()^〇11〇11 .Vector Predictor f0r shape; hereinafter referred to as Mvps; ^ 〇, in the present embodiment, the (MVPs) are represented by two-bit values, that is, (Mvp^ = (0, 0), and proceeds to step S208. In step S208, the block corresponding to Mvps is found in the reference video object plane 2, that is, the reference block 211, and the reference block 2ΐι and the desire are calculated. The sum of the absolute differences between the coding blocks 111 (Sum 〇f Ab(10)1 e e • Difference; hereinafter referred to as SAD), 胄 Describe, since the calculation method of sad is a technology of general video coding, and is not the technical feature of the case. Therefore, it will not be described here, and then proceeds to step S2〇9. In the step display, it is judged whether the SAD is smaller than the matching error idle value, and in the present embodiment, the matching error threshold is a standard value specified by the industry. That is (16*AlphaTH) 'If no, proceed _, if yes, proceed to step S211. In the step of the discussion, within the predetermined area range, that is, to listen to s = two and add, for example, 5 and 5 pixels as the radius of the region to search for The SAD between the blocks to be coded lu is the smallest block, = calculating the formula to calculate the final intestine of the block to be coded, and (4) 'the calculation formula is the shape movement vector difference value = (MVs) The processing step of moving the vector difference into the video encoding of the shape can be completed according to the obtained shape and ending the video encoding method of the present invention. 'In step S211, 'the shape movement of the block to be encoded m is set. 14 18769 (amendment) is equal to the body S, that is, the _ is taken as the shape movement vector of the current coded area 111, and the amount is moved as the meteorite, and the shape of the sundial is moved toward ::::= The motion estimation method of the video editing method of the present invention is that the motion estimation method of the present invention integrates each block of the conventional block into a block group of the domain according to the type, and the effective moving vector exists in the block group. ° _b To use the different counts of the ancient = weight block The shape motion vector predictor is compared with the first effective motion vector in the t-f vector block of the technical code, as the shape vector prediction I of the human, flat code block, and the estimation method of the present prediction value. More reasonable 1 private movement, because of the consumption (4) to the accuracy of the predicted value, the large and the shape of the motion vector predictor is not a small range to search for the block to be coded finally == 丨 can be good The purpose of the operational performance of the system is that the above description is only a preferred embodiment of the present invention and is limited to the scope of the present invention, that is, the present invention is used, for example, in the rhyme of the present invention. Only the other changes can still be made, and the preset range of the search described in the processing step S210 of the mobile estimation method is not limited to U>21〇 as the radius, that is, the end view implementation type : 5 and minus 5 pixel radius. Therefore, it is one of the spirits and technical thoughts of the present invention that is familiar with the number of == the same number of pixels. The disclosure of the present invention is covered by the scope of the patent application described later. , effect decoration or change 'still should be 18769 (Revised) 15 1283133 [Simple description of the diagram] Figures 1A and 1B are an example of the figure, after the complex encoding process for a video object, 糸 / knife frequency I don't know the information and texture information; 〇p) The shape generated by the second figure is used to illustrate the schematic diagram of the video motion cage 3Δ闰 total field θ private motion estimation; the brother 3Α diagram is used to show the present invention A schematic diagram of a first embodiment of a candidate vector block; and FIG. 4 and a corresponding third diagram thereof are used to display a second embodiment of the present invention for encoding blocks and their neighboring vector blocks; and 4 is a video editing stone of the invention ^ ^ ^ ^ ^ ^ • map. k (4) Estimating the operation of the village method [Key element symbol description] 11 110 :·111 113,115,117,119 ^ 12 120,122,124,126 123,125,127,129 2 211 S200~S211 To encode the video object plane shape information (Shape) shape Block (BAB) Precoding block neighbor block (shape block) Texture information (Texture) Texture block (Macro Block) Neighborhood block (texture block) Reference video object plane test block Step 16 18769 (corrected this)