[go: up one dir, main page]
More Web Proxy on the site http://driver.im/

TW200427332A - Rate control method with region of interesting support - Google Patents

Rate control method with region of interesting support Download PDF

Info

Publication number
TW200427332A
TW200427332A TW92113480A TW92113480A TW200427332A TW 200427332 A TW200427332 A TW 200427332A TW 92113480 A TW92113480 A TW 92113480A TW 92113480 A TW92113480 A TW 92113480A TW 200427332 A TW200427332 A TW 200427332A
Authority
TW
Taiwan
Prior art keywords
macro block
control method
image
important
patent application
Prior art date
Application number
TW92113480A
Other languages
Chinese (zh)
Other versions
TWI224469B (en
Inventor
Yung-Ching Chang
Original Assignee
Silicon Integrated Sys Corp
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 Silicon Integrated Sys Corp filed Critical Silicon Integrated Sys Corp
Priority to TW92113480A priority Critical patent/TWI224469B/en
Application granted granted Critical
Publication of TWI224469B publication Critical patent/TWI224469B/en
Publication of TW200427332A publication Critical patent/TW200427332A/en

Links

Landscapes

  • Compression Or Coding Systems Of Tv Signals (AREA)

Abstract

A rate control method with region of interesting support for coding macroblocks with different priority. The rate control method comprises the steps of: calculating a weighted macroblock activity according to the priority; calculating a picture activity according to the weighted macroblock activity; allocating a bit budget for each macroblock according to the priority; calculating an estimated complexity according to a complexity of a previously coded picture, an activity of the previously coded picture and the weighted macroblock activity; calculating an estimated quantizer scale according to the estimated complexity and the bit budget; calculating an initial virtual buffer occupancy according to an reaction factor and the estimated quantizer scale; calculating a macroblock quantizer scale according to a virtual buffer occupancy of a previously coded macroblock, the priority and the reaction factor; coding each macroblock according to the macroblock quantizer scale; and updating a virtual buffer occupancy of the current coded macroblock.

Description

200427332 五、發明說明(l) -------- 【發明所屬之技術領域】 本發明係關於壓縮率控制方法,特別是關於具有重 區域(jeg1〇n of interesting,R〇I)支援的壓縮率控制方 法,藉以提昇局部區域之影像品質,並有效控制使用位 元0 二、【先前技術】 為了縮小影像資料量,一般會對影像進行壓縮,例如 MPEG格式之影像編碼((:〇(^1^)。影像編碼之基本單元200427332 V. Description of the invention (l) -------- [Technical field to which the invention belongs] The present invention relates to a compression rate control method, and in particular, to support with a heavy region (jeg1〇n of interesting (RoI)). The compression rate control method is used to improve the image quality of the local area and effectively control the use of bit 0. [Previous technology] In order to reduce the amount of image data, the image is generally compressed, such as MPEG format image encoding ((: 〇 (^ 1 ^). The basic unit of image coding

(basic unit)為巨集區塊(macr〇_bi〇ck)。若取樣 (sampling)為4:2:0格式,則每個巨集區塊具有6個區塊 (block),分別為4個Y區塊、一個Cb區塊、以及一個^區 塊。每個區塊必須先經過離散餘弦轉換(discrete⑶以此 transform,以下簡稱DCT)後,再將DCT係數 (coefficient)量化成整數(integer)。一種鋸齒形 (zigzag)掃描或其他替代之掃描方法被用來將二維 (t w 〇 - d i m e n s i ο n s )之量化係數陣列排列成一維 (oneydimension)資料。最後,再利用所謂可變長度編碼 (variable-length coding,以下簡稱VLC)進行熵編碼 (entropy coding) 〇 通4 ’DCT係數F[v][u]的量化可表示成: QF[v][u] = 16 * F[v][u] / (q * w[v][u])".(1) 其中’ v、u為二維陣列之指標(index),範圍從〇〜7。 Q為每個巨集區塊内之各區塊的量化比例因子(quant izer(basic unit) is a macro block (macr_bioc). If the sampling is in the 4: 2: 0 format, each macro block has 6 blocks, which are 4 Y blocks, one Cb block, and one ^ block. Each block must undergo discrete cosine transform (hereinafter referred to as DCT), and then the DCT coefficient (coefficient) can be quantized into an integer. A zigzag scan or other alternative scanning method is used to arrange a two-dimensional (t w ω-di i m n s i ο n s) array of quantization coefficients into one-dimensional (oneydimension) data. Finally, the so-called variable-length coding (hereinafter referred to as VLC) is used for entropy coding. Quantization of 4'DCT coefficients F [v] [u] can be expressed as: QF [v] [ u] = 16 * F [v] [u] / (q * w [v] [u]) ". (1) where 'v and u are indexes of the two-dimensional array, ranging from 0 to 7 . Q is the quantization scale factor of each block in each macro block.

200427332200427332

scale),可隨著不同之巨集區塊改變QW[v][u]g整張圖 像(p1CtUre)所定義之量化矩陣(quantizati〇n , 用來指定每個不同DCT係數之權值(weighting fact〇r)。 圖1顯示在MPEG-2規格中對内部區塊(intra bl〇ck)所定義 的預,量化矩陣。如圖丨之量化矩陣所示,該量化矩陣的 ,隨著陣列指標v、u的增加而逐漸變大。此特性使得具有 咼頻之ϊ化係數較易變成〇,原因是人類的眼 信號較敏感,而對高頻信號較不敏感。 、低頻 在一般情形下,編碼器會將一張影像(image)視為單 一的物件(object)來編碼,因此整張影像之視覺品質 (visual quality)是單一性的(unif〇rm)。然而,在某些 情形下,我們希望圖像中的部分區域能以較好的品質進行 、、爲碼例如,監視系統(survei 1 lance system)即為一個 好的例子。一般的監視系統會有複數個攝影機連接到單一 錄影系統,從各個攝影機傳來的影像會被縮小並結合成單 一影像。然而,當我們希望某些攝影機比其他攝影機具 較高之影像品質時,傳統的錄影系統將無法滿足該要^。 另外一個例子是對於影像中混合標題(capti〇n)時之 資料流的編碼。由於標題一般均固定於某一位置,但曰σ北 景部不斷在移動,因此在標題周圍的環形假像(ringi =月 artifact)會變的很明顯。若能在該區域提供較多之 率,則该影像之整體視覺品質將可被提昇。那些我 獲得較好品質的區域被定義為重要區域(regi〇n 〇f interesting ,R0I)。scale), which can change the QW [v] [u] g whole image (p1CtUre) quantization matrix (quantizati0n) with different macroblocks to specify the weight of each different DCT coefficient ( weighting fact〇r). Figure 1 shows the pre- and quantization matrices defined for the intra block in the MPEG-2 specification. As shown in the quantization matrix in Figure 丨, the The index v and u increase gradually and become larger. This characteristic makes the frequency coefficient with high frequency easily become 0, because human eye signals are more sensitive, but less sensitive to high frequency signals. The encoder will encode an image as a single object, so the visual quality of the entire image is unifom. However, in some cases We hope that some areas in the image can be performed with better quality. For example, a surveillance system (survei 1 lance system) is a good example. A general surveillance system will have multiple cameras connected to a single recording. System from each camera The image will be reduced and combined into a single image. However, when we want some cameras to have higher image quality than other cameras, traditional video recording systems will not be able to meet this requirement. Another example is for mixed titles in images ( capti〇n) encoding of the data stream. Because the title is generally fixed at a certain location, but the northern part of σ is constantly moving, so the ringi (ringi = month artifact) around the title will become obvious If more rates can be provided in this area, the overall visual quality of the image will be improved. Those areas where I have obtained better quality are defined as important areas (regi0n 0f interesting, ROI).

200427332200427332

五、發明說明(3) 但疋’目前壓縮率控制方 碼控制。同時,— 並,又有支援重要區域之編 bi t rate)之限法丨丨 m ^位凡率(target 以達到使用者要求之重 :;集£塊之目標位元率 八之Ϊ要區域的效果將是重要問題。 三、【發明内容】 有鑒於上述問冑’本發明之目的是提供 J域且能分配適當位元預算值給各區域之壓縮率控=要 •為達成上述目的,本發明壓縮率控制方法包含下列舟 驟·计异出每個巨集區塊之活動程度Actj後,根據該巨 集區塊之優先權Pri一i計算該巨集區塊之權值活動程度 WAct —i,且根據每個巨集區塊之權值活動程度^以一丨^算 該圖像之活動程度Act — C ;分配該圖像之位元預算值 # ΒΒ一C ;分配每個巨集區塊之位元預算值ΒΒ—丨;計算估測複 雜度EstCom_C以及估測量化比例因子EstQ —C ;計算虛擬緩 衝佔用篁的初值D 0,計鼻每個巨集區塊之量化比例因子q 並編竭該巨集區塊;更新虛擬緩衝佔用量h ;判斷該圖像 是否已編碼完成’若已編碼完成則進行下一步驟,否則跳 回計算下一個巨集區塊之量化比例因子Qi+1並編碼該巨集區 塊之步驟;以及計算複雜度Com — X以及活動程度Act一X,作 為下一張圖像之參考。V. Description of the invention (3) But 疋 ′ is currently controlled by the compression rate control code. At the same time, — and, there is a limit method that supports the compilation of important areas 丨 丨 m ^ bit rate (target in order to achieve the weight of the user requirements: the target bit rate of the set of eight blocks is the most important area The effect of this will be an important issue. [Summary of the Invention] In view of the above question, the purpose of the present invention is to provide the J-domain and can allocate the appropriate bit budget value to the compression ratio control of each region. The compression rate control method of the present invention includes the following steps: After calculating the activity level Actj of each macro block, calculate the weight activity level WAct of the macro block according to the priority Pri_i of the macro block. —I, and the activity level of the image is calculated according to the weight value of each macroblock Act — C; the bit budget value of the image is allocated # ΒΒ 一 C; each macro is allocated Calculate the estimated complexity of EstCom_C and the estimated scale factor EstQ —C; calculate the initial value D 0 of the virtual buffer occupation ,, and calculate the quantized proportion of each macro block Factor q and edit the macro block; update the virtual buffer occupation h; Determine whether the image has been encoded 'If the encoding is completed, proceed to the next step, otherwise skip back to the step of calculating the quantization scale factor Qi + 1 of the next macro block and encode the macro block; and calculate the complexity Com — X and the degree of activity Act-X, as a reference for the next image.

第8頁 200427332 五、發明說明(4) 四、【實施方式】 以下參考圖式詳細說明本發明具有重要區域支援的壓 細率控制方法。 在MPEG編碼技術中,若巨集區塊以較小之量化比例因 子來進行量化時,其解碼後之的影像品質會較好。本發明 具有重要區域支援的壓縮率控制方法即利用該特性,指定 圖像中之部分巨集區塊具有較高之優先權(pri〇rity),並 以較小之i化比例因子來編碼這些巨集區塊。圖2所示為 一張圖像中巨集區塊具有不同優先權分布的優先權表之例 子。如該圖所示,每個格子代表一個巨集區塊,每個巨集 區塊可以個別指定一個優先權。例如,圖2所示中,橢圓 形區域被定義為重要區域,因此指定給該重要區域之巨集 區塊的優先權為2,其餘之巨集區塊的優先權為丨。圖3顯 示根據圖2之優先權表所轉換的量化比例因子表。如圖3'所 不,優先權指定為1之巨集區塊的量化比例因子為8, 權指定為2的巨集區塊,其量化比例因子為4,僅為優 指定為1之巨集區塊的量化比例因子的—半。所以,優先 權較高之巨集區塊可以得到較好的編碼影像品質。炎 根據編碼時的位元率限制,編碼器必須對每張圖 配一個位元預算值(bit budget)BB__c。之後,編碼器必 以適當之量化比例因子對每個巨集區塊進行編碼,萨以將 貫際使用的m率控制在#近位元預算值的範圍 i!:之具有重要區域支援的壓縮率控制方法,編碼器不 僅要控制所使用之位元率,還必須控制每個巨 z 200427332Page 8 200427332 V. Description of the invention (4) IV. [Embodiment] The method for controlling the compression ratio with important area support of the present invention will be described in detail below with reference to the drawings. In the MPEG coding technology, if the macroblock is quantized with a small quantization scale factor, the decoded image quality will be better. The compression rate control method with important region support of the present invention uses this feature to specify that certain macroblocks in an image have a higher priority, and encode these with a smaller scale factor. Macro block. Figure 2 shows an example of a priority list for macro blocks in an image with different priority distributions. As shown in the figure, each grid represents a macro block, and each macro block can be individually assigned a priority. For example, in Figure 2, the elliptical area is defined as an important area, so the priority of the macro block assigned to the important area is 2, and the priority of the remaining macro blocks is 丨. FIG. 3 shows a quantization scale factor table converted from the priority table of FIG. 2. As shown in Fig. 3 ', the quantization scale factor of the macro block designated as 1 is 8 and the macro block designated as 2 is quantized. The quantization scale factor is 4, which is only for the macro designated as 1. The quantization scale factor of the block-half. Therefore, the macro block with higher priority can get better coded image quality. Yan According to the bit rate limit during encoding, the encoder must assign a bit budget BB__c to each picture. After that, the encoder must encode each macro block with an appropriate quantization scale factor. Sae will control the m rate used throughout the #near bit budget range i !: It has compression that supports important regions. Rate control method, the encoder must not only control the bit rate used, but also each giant z 200427332

像品質。 在進彳丁編碼之前,pl£旦/ 須妯斗曾山士a 張衫像之活動程度(a c t i v i t y)必 叙铲许沾她人犯 像之活動程度為每個巨集區塊之活 掄而-# ^ ^ 個内〇M爲碼(intra-coded)巨集區 塊而έ ,其活動程度為個 · 曰, · 、 1 口似冗度(11111111131106)區塊之變異 里(variance)的總和·而祖士人 .. 而對於一個關聯性編碼 C inter 一 coded)巨集區换而山 林 也 _ ,^ 罘匕塊而吕,其活動程度為四個剩餘 (隹reSldUal)區塊之變異量(variance)的總和。之後,該巨 $ 塊之活動程度根據該巨集區塊之優先權&丨—丨來調整Like quality. Before entering the coding, the activity level of pl / dan / Zhudou Zengshanshi a shirt image must be described as the activity level of each of the macroblocks- # ^ ^ Inside 〇M is an intra-coded macro block, and its degree of activity is the sum of the variations in the block that is like redundant (11111111131106). And the ancestors: For a related code C inter a coded) macro area in exchange for the mountain forest _ ^ 罘 dagger block and Lu, its activity level is the variation of the four remaining (隹 reSldUal) blocks ( variance). After that, the activity level of the giant $ block is adjusted according to the priority of the macro block & 丨 — 丨

成權值活動程度WAct_i,亦即= Actj —* ρΗ —土。 而及圖像之活動矛壬度即為每個巨集區塊之權值活動程度 WAct — i的總和。在圖像編碼完成之後,平均量化比例因子 Avg_Q與實際使用位疋量⑽可以被計算出來,因此該圖像 之複雜度Com—C可根據式(1)計算出來:Weighted activity degree WAct_i, that is = Actj — * ρΗ — soil. And the activity of the image is the sum of the weight activity of each macroblock WAct — i. After the image coding is completed, the average quantization scale factor Avg_Q and the actual bit size ⑽ can be calculated, so the complexity of the image Com-C can be calculated according to formula (1):

Com—C = Avg—Q * UB …⑴Com—C = Avg—Q * UB… ⑴

、因此,在x—圖像(X可為I、P或B)編碼完後,該圖像之 活動程度Act —C與複雜度Com一C被用來更新X-圖像之活動程 度Act一X以及X-圖像之複雜度c〇m—x。所以,使用上一張X一 圖像之活動程度Act —X、複雜度Com一X、以及目前圖像之活 動程度Act— C即可計算出目前圖像的估測複雜度 EstCom_C。其計算方式如式(2)所示:Therefore, after the x-image (X can be I, P, or B) is encoded, the activity level of the image Act-C and the complexity Com-C are used to update the activity level of the X-image Act- The complexity of X and X-images is com-x. Therefore, the estimated complexity EstCom_C of the current image can be calculated by using the activity level Act-X, the complexity Com-X, and the current activity level Act-C of the previous X-image. The calculation method is shown in equation (2):

EstCom—C = Act—C * Com—X/Act_X."(2) 而且’利用目前圖像的估測複雜度EstCom — C以及目前 圖像之位元預算值BB一C,即可計算出目前圖像的估測量化EstCom—C = Act—C * Com—X / Act_X. &Quot; (2) And 'Using the estimated complexity of the current image EstCom — C and the bit budget value BB-C of the current image, we can calculate Estimation of current images

第10頁 200427332 五、發明說明(6) 比例因子E s t Q —C。其計算方式如式(3 )所示:Page 10 200427332 V. Description of the invention (6) Scale factor E s t Q —C. The calculation method is shown in formula (3):

EstQ_C = EstCom_C / BB—C …(3) 估測量化比例因子EstQ一C為優先權1之巨集區塊的預 設量化比例因子。所以,若巨集區塊之優先權為PrLi, 則該巨集區塊之量化比例因子應為—C/pr i i。所以, 巨集區塊之優先權越高,其量化比例因子越小。EstQ_C = EstCom_C / BB-C… (3) Estimation and measurement scale factor EstQ-C is a preset quantization scale factor for the macro block of priority 1. Therefore, if the priority of the macro block is PrLi, the quantization scale factor of the macro block should be -C / pr i i. Therefore, the higher the priority of a macroblock, the smaller its quantization scale factor.

然而,估測量化比例因子E st Q一C並非完全正確,且若 編碼器以固定之估測量化比例因子— C來編碼圖像,則 貫際使用之位元與位元預异值將會產生較大差距。所以一 種稱為虛擬緩衝(virtual buffer)的回授機制(feedback mechanism)被用來控制位元的消耗(bit consumptiQn)。 R一X被定義為X-圖像之反應參數(reacti〇n faetQr^。 一般情形下,反應參數R —X被設定成2*位元率(bit rate)/ 圖框率(frame —rate)。在圖像編碼之前,虛擬緩衝佔用量 的初值DO被設定為R —X/31*EstQ —C,且位元預算值Ββ c根里 據母個巨集區塊之權值活動程度f A c t — i被事先户定认— 巨集區塊。所以,編碼器在對巨集區塊仙—i進行編碼時, / Pr i —i ··· (4)However, the estimated scale factor E st Q-C is not completely correct, and if the encoder encodes the image with a fixed estimated scale factor-C, the bits and bit pre-differential values that are used consistently will be Create a large gap. Therefore, a feedback mechanism called a virtual buffer is used to control the bit consumptiQn. R-X is defined as the reaction parameter of X-image (reactioon faetQr ^. In general, the reaction parameter R-X is set to 2 * bit rate / frame rate Before the image encoding, the initial value DO of the virtual buffer occupancy is set to R —X / 31 * EstQ —C, and the bit budget value Bβ c is based on the weight activity degree f of the macro block. A ct — i is identified in advance by the user — macro block. Therefore, when the encoder encodes the macro block fairy — i, / Pr i — i ··· (4)

Dy * 31Dy * 31

其中,為上一巨集區塊編碼完後之虛擬緩衝佔 量。而在編碼完每一張圖像後,編碼器係根據該巨 ^ 事先所分配的位元預算值BB_i以及該巨集區塊^ 11 元里UB__i來更新虛擬緩衝佔用量])。其更新A 用$ : 4万式如式(5 7L 示Among them, the virtual buffer share after the encoding of the previous macro block. After encoding each image, the encoder updates the virtual buffer occupancy according to the bit budget value BB_i allocated in advance by the macro ^ and UB__i in the macro block ^ 11 yuan]). Its update A uses $: 40,000 formulas (5 7L shown

200427332 五、發明說明(7)200427332 V. Description of Invention (7)

Di = Dh + UB—i - BB—i …(5) 所以丄當該巨集區塊已使用之位元量卯一丨大於所分配的位 ,預异值BB—i,虛擬緩衝佔用量Di就會增加且下一個巨集 區塊之量化比例因子亦相對提高。相反的,當該巨集區塊 ^使用之位元量UB—i低於所分配的位元預算值BBj,虛擬 緩衝佔用量Di就會降低且下一個巨集區塊之量化比例因子 亦,對降低。所以,當編碼器對該圖像編碼完成後,即可 $得平均量化比例因子AVg一Q以及所使用位元量UB — C,同 時該圖像之複雜度亦可被計算出來。 &圖4顯示本發明具有重要區域支援的壓縮率控制方法 *流程圖。如該圖所示,本發明具有重要區域支援的壓縮 率控制方法包含下列步驟: 步驟S402 ·計算每個巨集區塊之權值活動程度WActj >ΐ之活動程度Act-C。亦即,計算出每個巨集區塊 兮I^i後,根據該巨集區塊之優先權計算 厂區塊之權值活動程度WAct-i,亦即WAct —i = Act —i 隼指標土為第土個巨集區塊。接著,將每個巨 動程度WACt」加總後作為目前圖像之活動 目义=S4〇4 ·分配每個巨集區塊之位元預算值BB—i。將 程二A ?元預算值BB-C根據每個巨集區塊之權值活動 又c -1來分配每個巨集區塊之位元預算值。 例因·計算估測複雜度Estc°m-c以及估測量化比 s -。編碼器根據目前圖像之活動程度Act-C、Di = Dh + UB—i-BB—i… (5) So, when the amount of bits used in the macro block is larger than the allocated bits, the pre-differential value BB—i, the virtual buffer occupation Di It will increase and the quantization scale factor of the next macro block will also increase relatively. Conversely, when the bit amount UB-i used by the macro block ^ is lower than the allocated bit budget value BBj, the virtual buffer occupancy Di will decrease and the quantization scale factor of the next macro block will also, Right down. Therefore, after the encoder finishes encoding the image, the average quantization scale factor AVg-Q and the amount of bits UB-C used can be obtained, and the complexity of the image can also be calculated. & FIG. 4 shows a flow chart of a compression rate control method with important area support according to the present invention. As shown in the figure, the compression rate control method with important area support of the present invention includes the following steps: Step S402-Calculate the activity level Act-C of the weight activity level WActj > of each macroblock. That is, after calculating each macro block I ^ i, the weight activity level WAct-i of the factory block is calculated according to the priority of the macro block, that is, WAct —i = Act —i 隼 indicator The soil is the first macro block. Next, the sum of each macro-motion degree WACt ”is taken as the activity of the current image. Objective = S404. Allocate the bit budget value BB-i of each macro block. The budget value BB-C of Cheng II will allocate the bit budget value of each macro block according to the weight activity of each macro block and c -1. Example: Calculate estimated complexity Estc ° m-c and estimate measured ratio s-. The encoder is based on the current activity level of the image Act-C,

第12頁 200427332 五、發明綱⑻ " 刖一張圖像之活動程度Act —X以及前一張圖像之複雜度 Com一X來計算出目前圖像之估測複雜度Estc〇m — c,其計算 方式如上述之式(2 )所示。接著,編碼器根據目前圖像之 ,測複雜度EstCom—C以及目前圖像之位元預算值BB__C來計 算出目前圖像之估測量化比例因子EstQ — c,其計算方 上述之式(3 )所示。 步驟S40 8 · #异虛擬緩衝佔用量的初值DQ。編碼器根 一反應參數R — X以及目前圖像之估測量化比例因子 來計算虛擬緩衝佔用量的初值⑽,該虛擬緩衝佔用 里的初值DO被設定為R_x/31*EstQ —c 〇 步驟 S 41 0 :計算备 ^fiii s r〇 ^ ,, 碼該巨隹F换姑集£塊罝比例因子Qi並編 場椒f 編碼器根據前一個編碼之巨集區塊的#擬 綾衝佔用量D· ,、/5旛姿机D v 求匕呢的虛擬 個巨隼反應參數1^、以及優先權Pri_i來計算每 所^ 例因子Qi,其計算方式如上述之式^ 碼器即ii:;: 巨集區塊之量化比例因子a後,編 2例因子Q1對該巨集區塊進行編瑪。 集區塊編碼完後更量D;:在編碼器對-巨 虛擬緩衝佔用量])X據刖個編碼之巨集區塊的 該巨集區塊之已傕1用彳^集區塊之位元預算值、以及 i ’其計瞀方, 70值⑽-1來更新虛擬緩衝佔用詈])Page 12 200427332 V. Outline of the invention " 活动 The degree of activity of an image Act —X and the complexity of the previous image Com — X to calculate the estimated complexity of the current image Estc〇m — c The calculation method is shown in the above formula (2). Then, the encoder calculates the estimated scaling factor EstQ — c of the current image based on the current image, the measured complexity EstCom—C, and the bit budget value BB__C of the current image, which is calculated by the above formula (3 ). Step S40 8 # The initial value DQ of the different virtual buffer occupancy. The encoder uses a response parameter R — X and the estimated scale factor of the current image to calculate the initial value of the virtual buffer occupancy ⑽. The initial value DO in the virtual buffer occupancy is set to R_x / 31 * EstQ —c 〇 Step S 41 0: Calculate the backup ^ fiii sr〇 ^, and code the giant F to change the set of blocks £ the scale factor Qi and edit the field pepper f encoder according to the # of the macro block previously encoded Quantities D · ,, / 5 pose machine D v find the response parameter 1 ^ and the priority Pri_i of each virtual giant to calculate the factor Qi for each example. The calculation method is as the above formula. :;: After the quantization scale factor a of the macro block, edit the factor Q1 of the macro block to edit the macro block. After the block encoding is completed, the amount is D ;: in the encoder pair-the giant virtual buffer occupancy]) X According to one encoded macro block, the macro block has been used for 1 block. Bit budget value, and i 'its calculation method, 70 value ⑽-1 to update the virtual buffer occupation 詈])

J二方式如上述之式⑸所示。 …占用里D 所有的巨集區塊繼續進行編竭的動作。ΐ 疋3已編碼元成,則跳至下一步驟。The second method is shown in the above formula (i). … Occupy all the macro blocks in D and continue to perform the depletion action. ΐ 疋 3 has been encoded, then skip to the next step.

第13頁 區塊i未』^ :判斷該圖像是否已編碼完成。若尚有P隹 200427332 五、發明說明(9) 步驟S416 :更新複雜度c〇m_X以及活動程度Act一X。當 該圖像已編碼完成後,編碼器根據平均量化比例因子 Avg — Q以及該圖像之已使用位元量ub_C來更新複雜度 Com一X ’其計异方式為c〇m_x = Avg —Q * UB —C。同時’將 目前圖像之活動程度Act_C指定為Act_X。 由於本發明之具有重要區域支援的壓縮率控制方法’ 對於優先權較高之巨集區塊以較小之量化比例因子進行編 碼,因此,可以得到較好的編碼影像品質。 以上雖以實施例說明本發明,但並不因此限定本發明 之範圍,只要不脫離本發明之要旨,該行業者可進行各種 變形或變更。Page 13 Block i ^^: Determine whether the image has been encoded. If there is still P 隹 200427332 V. Description of the invention (9) Step S416: Update the complexity com_X and the activity level Act-X. After the image has been encoded, the encoder updates the complexity Com_X 'according to the average quantization scale factor Avg — Q and the used bit quantity ub_C of the image. The calculation method is c0m_x = Avg —Q * UB —C. At the same time, the activity level of the current image, Act_C, is designated as Act_X. Since the compression rate control method with important region support of the present invention 'encodes a macro block with a higher priority with a smaller quantization scale factor, a better coded image quality can be obtained. Although the present invention has been described by way of examples, the scope of the present invention is not limited thereby, and those skilled in the art can make various modifications or changes without departing from the gist of the present invention.

第14頁 §37 200427332 圖式簡單說明 圖1顯示在MPEG-2規格中對内部區塊所定義的預設量化 矩陣。 圖2所示為一張圖像中巨集區塊具有不同優先權分布的 優先權表之例子。 圖3顯示係根據圖2之優先權表所轉換的量化比例因子 表。 圖4顯示本發明具有重要區域支援的壓縮率控制方法的 流程圖。Page 14 §37 200427332 Brief description of the figure Figure 1 shows the preset quantization matrix defined for the internal block in the MPEG-2 specification. Figure 2 shows an example of a priority list for macro blocks in an image with different priority distributions. FIG. 3 shows a quantization scale factor table converted from the priority table of FIG. 2. Fig. 4 shows a flowchart of a compression rate control method with important area support according to the present invention.

Claims (1)

200427332200427332 200427332 六、申請專利範圍 比例因子(¾ 編碼該巨集區塊,根據前述量化比例因子Q編碼該巨 集區塊; 更新虛擬緩衝佔用量Di,係根據該巨集區塊之已使用 位凡UB—i、上一個已編碼巨集區塊之虛擬緩衝佔用量a」、 以及該巨集區塊之位元預算值BB—i來更新虛擬緩衝佔用量 Di ;以及 判斷該圖像是否已編碼完成,若已編碼完成則進行下 一步驟,否則跳回計算下一個巨集區塊之量化比例因子Q i + l 2 ·如申明專利範圍第1項所記載之具有重要區域支援 的壓縮f控制方法,還包含下列步驟: 、°十二複雜度C〇m-X,係根據目前圖像之使用位元UB_C 以及目珂、圖像之平均量化比例因子Avg—Q計算複雜度 V/ Ο Π1 _ A,iJs r产aU?,度杬^ ’係將編碼完之目前圖像的活動 矛王度八(31: —C扣疋給活動程度Act —X。 的壓3二二:m ΐ2項所記載之具有重要區域支援 ^ ^ - «Act. , , „ 的壓範^二項所記載之具有重要區域支援 為該目前圖像之每個;隼D目:圖像之活動程度Act-C 和。 母個巨集區塊之權值活動程度WAct_i的總200427332 VI. Patent application scale factor (¾ encodes the macro block, encodes the macro block according to the aforementioned quantization scale factor Q; updates the virtual buffer occupancy Di, based on the used block UB of the macro block —I, the virtual buffer occupancy a of the last coded macro block, and the bit budget value BB-i of the macro block to update the virtual buffer occupancy Di; and determine whether the image has been encoded If the encoding is completed, proceed to the next step, otherwise skip back to calculate the quantization scale factor Q i + l 2 of the next macro block. · Compression f control method with important area support as stated in the first patent scope , Also includes the following steps: °, twelve complexity ComX, based on the current image using the bit UB_C and the average quantization scale factor Avg-Q of the image, image V / Π1 _ A, iJs r produces aU ?, degree 杬 ^ 'is the active spear king of the current image after encoding Du Duba (31: -C buckle to the degree of activity Act -X. The pressure 32: 22: m ΐ 2 With important regional support ^ ^-«Act.,, „Demonstration ^ The important areas recorded in the two items are supported for each of the current images; 隼 D: the activity level of the image Act-C and. The activity level of the weight of the parent macro block WAct_i total 第17頁 200427332 六、申請專利範圍 的Λ ΐ申請專利範圍第2項所記載之具有重要區域支援 制方法 EstC0m—c = Act—C * Com—X/Actj 。 ^申請專利範圍第5項所記載 的壓縮率控制方法,豆中 巧里文匕A叉板 子“切―C為: /、中别述目别圖像之估測量化比例因 EstQ_C = EstCom_C / bb—C 。 7 ·如申睛專利範圍第 的壓縮率控制方法,A中、=載之具有重要區域支援 為: 八則逑虛擬緩衝佔用量的初值DO D0 = R—X/31*EstQ—C 。 8·如申請專利範圍第7項 的壓縮率控制方法,苴中_、、斤5己載之具有重要區域支援 為·· ’、則述巨集區塊之量化比例因子A Qi = L * 31 / 9·如申請專利範圍第8 一 的壓縮率控制方法,其 貝所§己載之具有重要區域支援 h = Dh + ubj: 前述虛擬緩衝佔用量^為: 1 〇 ·如申請專利範圍第^1 的壓縮率控制方法,其中=項所記載之具有重要區域支援 Comj = Avg—Q =述複雜度C〇mj為: u·如申請專利範圍第。 的壓縮率控制方法,其中卷j所記載之具有重要區域支援 句4述巨集區塊為内部編碼之巨 IH 苐18頁 200427332 六、申請專利範圍 集區塊時,該巨集區塊之活動程度A c t _ i為四個亮度區塊 之變異量的總和。 1 2.如申請專利範圍第1項所記載之具有重要區域支援 的壓縮率控制方法,其中當前述巨集區塊為關聯性編碼之 巨集區塊時,該巨集區塊之活動程度Act_i為四個剩餘區 塊之變異量的總和。 1 3.如申請專利範圍第1項所記載之具有重要區域支援 的壓縮率控制方法,其中巨集區塊之優先權P r i _ i越大, 表示該巨集區塊越重要。Page 17 200427332 VI. Patent Application Λ ΐ The important regional support system described in Item 2 of the patent application scope EstC0m-c = Act-C * Com-X / Actj. ^ The compression rate control method described in item 5 of the scope of the patent application, Douzhong Qiao Liwen A fork board "Cut-C is: / 、 The estimated measurement ratio of the image in the catalogue in China is due to EstQ_C = EstCom_C / bb —C. 7 · If the compression rate control method in the patent scope of Shenyan, the support of important areas in A and = are as follows: The initial value of the virtual buffer occupancy DO D0 = R—X / 31 * EstQ— C. 8. If the compression rate control method of item 7 in the scope of the patent application, the important area support of 苴 中, 、, 斤 5 is set as ..., then the quantization scale factor of the macroblock A Qi = L * 31/9 · If the compression ratio control method of the 8th in the scope of patent application, it has important area support h = Dh + ubj: The above virtual buffer occupancy ^ is: 1 〇 · If the scope of patent application Compression ratio control method of item ^ 1, where == the important area support described in the item Comj = Avg—Q = the complexity C0mj is: u · As described in the scope of patent application, the compression ratio control method, where the volume j The recorded macroblocks with important regions are described in the macro block as internal compilation. Zhiju IH 苐 18 pages 200427332 VI. When applying for a patent range set block, the activity level A ct _ i of the macro block is the sum of the variations of the four brightness blocks. The compression rate control method with important area support described in the above item, wherein when the aforementioned macro block is a macroblock with correlation coding, the activity level of the macro block Act_i is the variation of the four remaining blocks 1 3. The compression rate control method with important area support as described in item 1 of the scope of patent application, wherein the larger the priority of the macro block P ri _ i is, the more important the macro block is. 第19頁Page 19
TW92113480A 2003-05-19 2003-05-19 Rate control method with region of interesting support TWI224469B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
TW92113480A TWI224469B (en) 2003-05-19 2003-05-19 Rate control method with region of interesting support

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
TW92113480A TWI224469B (en) 2003-05-19 2003-05-19 Rate control method with region of interesting support

Publications (2)

Publication Number Publication Date
TWI224469B TWI224469B (en) 2004-11-21
TW200427332A true TW200427332A (en) 2004-12-01

Family

ID=34568360

Family Applications (1)

Application Number Title Priority Date Filing Date
TW92113480A TWI224469B (en) 2003-05-19 2003-05-19 Rate control method with region of interesting support

Country Status (1)

Country Link
TW (1) TWI224469B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102263943A (en) * 2010-05-25 2011-11-30 财团法人工业技术研究院 Video bit rate control device and method
TWI685247B (en) * 2017-06-14 2020-02-11 晶睿通訊股份有限公司 Image processing method and related camera device and camera system

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102263943A (en) * 2010-05-25 2011-11-30 财团法人工业技术研究院 Video bit rate control device and method
CN102263943B (en) * 2010-05-25 2014-06-04 财团法人工业技术研究院 Video bit rate control device and method
TWI685247B (en) * 2017-06-14 2020-02-11 晶睿通訊股份有限公司 Image processing method and related camera device and camera system
US11403762B2 (en) 2017-06-14 2022-08-02 Vivotek Inc. Image processing method and related camera device

Also Published As

Publication number Publication date
TWI224469B (en) 2004-11-21

Similar Documents

Publication Publication Date Title
CN100362863C (en) Method and apparatus for selecting macroblock quantization in a video encoder
US8175149B2 (en) Method and apparatus for controlling bitrate of scalable video stream
CN103918262B (en) Code rate distortion optimization based on structural similarity perceives method for video coding and system
TWI304297B (en) Video encoding methods and systems with frame-layer rate control
CN1910934B (en) Encoder with adaptive rate control
US8254702B2 (en) Image compression method and image processing apparatus
TW571589B (en) Method of performing video encoding rate control using bit budget
US7194031B2 (en) Rate control method with region of interesting support
US8406297B2 (en) System and method for bit-allocation in video coding
Hu et al. Region-based rate control for H. 264/AVC for low bit-rate applications
CN110740318A (en) Automatic adaptive long-term reference frame selection for video processing and video coding
WO2004010702A1 (en) A bit-rate control method and device combined with rate-distortion optimization
CN103636188B (en) Encoder-supervised imaging for video cameras
CN101094411A (en) Code rate control method of video code
CN106060539B (en) A kind of method for video coding of low transmission bandwidth
JPH0974565A (en) Controller for video signal coding system
CN107211145A (en) The almost video recompression of virtually lossless
CN112188208B (en) Macro block level code rate control method and related device
TWI247546B (en) A video encoding method which carries out the encoding of P frame or B frame by utilizing I frame
CN103237221B (en) A kind of H.264 frame layer rate control method of structure based likeness coefficient
TW200427332A (en) Rate control method with region of interesting support
WO2008018708A1 (en) Image quantization method and apparatus
TW200537943A (en) Method and apparatus for compressing video data
KR20100032619A (en) System and methods for rate control based model parameter, storage medium storing the same
Kunter et al. Extending H. 264/AVC with a background sprite prediction mode

Legal Events

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