200908758 玖、發明說明： 【發明所屬之技術領域】 本發明提供一種影像位移偵 PAL影像系統之影像位移偵測方法。，、彳日—種應用於 【先前技術】 影像資料於傳輸時，一種# .. 見的傳輸方法係藉由一合 成訊唬（composite Signal)加以值於 二人 ^ ^ ^π · 而0成訊號則攜帶有 冗度貝訊（1Umlnance lnf〇rmati〇n)以及彩产 (chr〇mi„ance information) ^ ^ ^ f IR ^ . 〇, 附加於亮度資訊的高頻訊號上。 式 在接收到合成訊號之後，影像解碼器㈣如如 必須將合成訊號所攜帶的亮度資訊與彩度資訊分離出 來，以進行後續的影像處理，而這樣的操作 = 彩度分離（Y/C分離）。 免度 而Y/C分離大致可以三種不同的操作方式 二維的Y/C分離、三維的Y/c分離、 、引馮 刀離、以及介於兩者 :/c刀離’其中’ 一維的Y/c分離較適合用於移動象 中，三維的Y/C分離則較適合用於靜止的影像中，、二像 最佳的影像品質。 ’以得到 由上述可知，若影像的位移狀況不 吓休用的Y/r 分離方式亦有所不同；因此，影像的位移仙 影像處理的一個重要關鍵，舉例來說，若偵測到的A成為 移狀況有誤，那麼便會於其後採用錯誤的γ 衫像位 , L刀離方彳 如此便可能會導致影像品質低落。 巧’ 200908758 如業界所習知，在NTSC影像系統中，在訊號傳輸時， 會先將RGB影像訊號轉換成γυν影像訊號，並且將彩度 資讯與梵度資訊調變成一合成訊號（如前所述，彩度訊號係 附加於亮度訊號的高頻部份），而合成訊號可以下列方程式 表示之：200908758 玖, invention description: [Technical Field] The present invention provides an image displacement detection method for an image displacement detection PAL image system. , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , The signal carries the ubiquitous information (1Umlnance lnf〇rmati〇n) and the color production (chr〇mi„ance information) ^ ^ ^ f IR ^ . 〇, attached to the high-frequency signal of the brightness information. After synthesizing the signal, the image decoder (4) must separate the luminance information and the chroma information carried by the synthesized signal for subsequent image processing, and such operation = chroma separation (Y/C separation). The Y/C separation can be roughly divided into three different modes of operation: two-dimensional Y/C separation, three-dimensional Y/c separation, von knive separation, and somewhere in between: /c knife away from 'where' one-dimensional Y The /c separation is more suitable for moving images. The three-dimensional Y/C separation is more suitable for use in still images, and the two images are optimal for image quality. 'According to the above, if the displacement of the image is not scared The Y/r separation method for rest is also different; therefore, the image An important key to the image processing of the displacement fairy. For example, if the detected A is in the wrong state of shifting, then the wrong gamma-shirt position will be used thereafter, and the L-knife will cause the image to be caused. The quality is low. Qiao' 200908758 As is well known in the industry, in the NTSC imaging system, when the signal is transmitted, the RGB image signal is first converted into a γυν image signal, and the chroma information and the Brahman information are converted into a composite signal. (As mentioned above, the chroma signal is attached to the high frequency portion of the luminance signal), and the composite signal can be expressed by the following equation:
Composite signal- Y+Usin(wt)+Vc〇s(wt) 方程式（1) . 其中¥為凴度訊號，而U、V為彩度訊號。Composite signal- Y+Usin(wt)+Vc〇s(wt) Equation (1) . where ¥ is the twist signal and U and V are the chroma signals.
. 由方程式（1)可知，彩度訊號上面有載波，而且NTSC f . 影像系統中定義上下兩條掃描線的載波反相（相位相差 1 80度）；此外，由於NTSC系統中，每個圖框具有525條 掃描線，因此相鄰兩圖框的同一條掃描線之間亦具有i 8〇 度的相位差。換句話說，在NTSC影像系統中，第n張圖 框的第m條掃描線與第n+2張圖框的第m條掃描線具有 相同的彩度載波相位；而實際上，只要兩圖框之間相隔奇 數個圖框，那麼此兩圖框的同一條掃描線之間之彩度載波 的相位便會相同。 理論上，於彩度載波相位相同的兩圖框中，其内部像 素的像素值所受到彩度載波相位的影響是相同的；換古 之，若影像是屬於一個靜止影像，理論上具有相同相位的 • 圖框會具有完全相同的像素值。因此，於NTSC系統中， ' 便可以利用第n張圖框與第n + 2張圖框的資訊來進行影像 位移之偵測。 但是，以PAL系統而言，合成訊號係以下列方程式表 不之 . 方程式（2)From equation (1), there is a carrier on the chroma signal, and the NTSC f. image system defines the carrier inversion of the upper and lower scan lines (phase difference of 1 80 degrees); in addition, because of the NTSC system, each figure The frame has 525 scan lines, so the same scan line between adjacent frames also has a phase difference of i 8 degrees. In other words, in the NTSC image system, the mth scan line of the nth frame has the same chroma carrier phase as the mth scan line of the n+2th frame; If the frames are separated by an odd number of frames, the phase of the chroma carrier between the same scan lines of the two frames will be the same. Theoretically, in the two frames with the same phase of the chroma carrier, the pixel values of the internal pixels are affected by the phase of the chroma carrier. In other words, if the image belongs to a still image, the theoretical phase has the same phase. • The frame will have exactly the same pixel value. Therefore, in the NTSC system, 'the information of the nth frame and the n + 2 frame can be used to detect the image displacement. However, in the case of the PAL system, the synthesized signal is not expressed by the following equation. Equation (2)
Composite signal^ Y+Usin(wt)±Vcos(wt) 200908758 由上述方程式（2)可知，PAL影像系統係定義上下兩條 水平掃描線的載波相差90度，換言之，第n條掃描線與 第η+2條掃描線的載波僅相差〗8〇度，而由於pal系統 中，每個圖框具有62 5條掃描線，因此前後兩圖框的同一 掃描線亦僅相差90度，亦即第n個圖框的第婦描線 與第n+2個圖框的第爪條掃描線的載波僅相差18〇度。換 言之，於PAL系統中，如果要利用具有才目同相位彩度載波 的兩圖框來進行影像位移偵測，便必須利用第η張圖框與 第η+4張圖框的資訊才能完成。 、 然而，這樣的做法在NTSC系統之中較不為難 田於 TSC系統僅須暫存兩張圖框，例如第n張圖框與第州 =圖框’再加上第n+2 m圖框巾數條掃描線之合成訊號資 L ’即可進行影像的位移偵測’因此對於緩衝器㈣㈣ =容量需求並不會過大；但是，纟PAL系統中，若要以相 ::法進行影像的位移偵測時，便需要暫存四張圖框， j如第η張圖框至第n+3張圖 ^ , 取圑框，再加上第n+4張圖框中 數條知描線之合成訊號資料， 糸 始侍進仃。很明顯地，在PAL· 糸、·先中’緩衝器的容量需求係$ 此，前述的做法並不經濟。 SC糸統的兩倍，因 【發明内容】 因此本發明之主要目的夕 偵測方# 在於提供一種影像位移 頂碉方法，其可於pAL影像丰 像位移彳貞^目^+利用較少的圖框達到影 求量太大的問題。 ▲知技術中緩衝器的容量需 200908758 根據本發明之一實施例，係揭露一種影像位移偵測方 法’其係應用於一 PAL影像系統，該影像位移偵測方法包 含有：接收一合成訊號（composite signal)，該合成訊穿攜 帶有一目標圖框（frame)以及鄰近於該目標圖框之至少— 第一參考圖框；以及根據該目標圖框所攜帶之_第—資 訊與該第一參考圖框所攜帶之一第一對應資訊，來判斷= 於該目標圖框之一目標像素之位移狀況；其中該第一資气 與該第一對應資訊具有相同之彩度載波相位。 / \ r \ 根據本發明之一實施例，另揭露一種影像位移偵測方 法，其係應用於一 PAL影像系統，該影像位移偵測方法包 含有：接收一合成訊號（composite signal)，該合成訊號攜 帶有—目標圖框（frame)以及鄰近於該目標圖框之一第— 參考圖框與一第二參考圖框；對於該目標圖框之一目標 =素，取得對應該目標像素之一影像區域，該影像區域: 2有複數個像素；對於該複數個像素中每一像素，從該第 =參考圖框或該第二參考圖框取得具有相同彩度相位之 —對應像素；算該複數個像素之每—㈣與其對應像 素f像素值差量，以取得複數個像素值差量；對該像素值 差量進行-加權平均運算，以取得—加權平均值；以及根 據該加權平均值，判斷該目標像素之位移狀況。 、根據本發明之一實施命卜另才焉露一種影像位移债測方 :，其係應用於一 PAL影像系統，該影像位移偵測方法包 含有：接收一合成訊號（composite signal)，該合成訊號攜 帶有一目標圖框（frame)以及鄰近於該目標圖框之一第_ 9 200908758 參考圖框與一第二參考圖框；對一 Ί、 a知像素，取得對應 δ亥目標像素之複數個參考像辛；稱播兮口 n n根據β亥目標圖框所攜帶 之一第一資訊與該第一參考圖框所攜帶之—第一對應次 訊，以及根據該目標圖框所攜帶之一第二資訊料第= 考圖框所攜帶之-第二對應資訊，來判斷該複數個料之 參考位移狀況；㈣康該複數個像素之參考位移狀況，判斷 該目標像素之位移狀況；其中該第一資訊與該第—對應 資訊具有相同之彩度載波相位，以及該第二資訊與該第二 對應資訊具有相同之彩度載波相位。 / 一 本發明之影像偵測方法於PAL影像系統中進行影像 位移偵測時，無須暫存四個圖框，而只須 πI瞀孖—至兩個 圖框即可達成位移债測的㈣，因Α，本#明對於緩衝器 (buffer)的容量需求較小，可以節省更多的成本。 【實施方式】 在此請參閱第1圖，第1圖為PAL影像系統中不同圖 框（N-2〜N+2)之同一掃描線μ的像素值示意圖。於第i圖 中’係假設每一條掃描線係取樣於載波的45、丨3 5、225、 3 15度，此外，為了簡化起見，每一條掃描線僅緣示出四 個像素，並且於各像素的旁邊亦標明了其像素值，特別是 其載波之相位關係。 在此請注意，於第1圖中’每個圖框N_2〜圖框n+2 的像素皆彼此對應’換言之，以下各級像素：（pi、p5、 P9、P13、P17) ; ( P2、P6、P10、P14、P18) ; ( P3、P7、 Pll、P15、P19) ; (P4、P8、P12、P16、P20)，係分別對 200908758 至圖框n+2上的相同位置。此外’在每-個像 應的像素值（譬如像素pi的像素值Yi+(V1+U1))之 彩度值代表了錢素的亮度值，巾V、U則代表該像素的 由第1圖可知，由於兩個相鄰圖框的載波相位相差90 又因此，其中載波相位完全相同的只有圖框&盘圖框 - 然而，兩相鄰圖框之間仍有對應的關係。以圖框Μ . 肖圖框心列’雖然以整條掃描線而言，彼此並沒有完全 同：，但是其中像素P5與像素p9、像素p7與像素⑴ 卻是同相的H面，以圖框N+1與圖框N為例，雖然 以整條掃描線而言，彼此並沒有完全同相，但是其中像素 P1〇與像素P14、像素P12與像素p16卻是同相的。 實際上，圖框N的一條掃描線之中，會有一半的像素 可以於前一個圖框N-1中找到同相的對應像素，而另一半 的像素則可於後-個圖框N+1中找到同相的對應像素。舉 例來說，以圖框N為例，一半的像素p9、pu分別盥前一 ®框叫的對應像素P5、P7同相，而另一半的像素ρι〇、 P12則與後一圖框N+1的對應像素pu、pi6同相。而於 圖框N中，與前一圖框之對應像素同相之該一半像素 以及與後一圖框N+1之對應像素同相之該另一半像素則 係相互交錯配置。 ' 如則所述，影像位移偵測的操作如欲得到最佳的效 果，則須使用到不同圖框中載波相位相同的對應像素，而 於習知技術中，這樣的性質有賴於參考每一對應像素之載 11 200908758 二目位::::兩:框(譬如前述的圖框N與圖框N+4)， 圖框中 ㈣前料性f，直接«框N之相鄰 圖框中，例如固4广、τ , ^ W州 灼如圖框Ν-1及圖框N+1中， 的對嶋，來㈣㈣㈣^取㈣波相位相同 用圖：5N:對於圖框U的一目標像素，本發明僅須利 =目:相：圖框—或一，便可達成位移 用兩=二前述可知，本發明可以取代習知技術中使 明在進： 框Ν與圖框Ν+4的做法。因此，本發 二=移侦測的時候，僅須暫存兩張圖框 暫存量，進而節省了緩二t 省了一半的資料 即噌了緩衝态的建置成本。 在此請參閲第2圖，筮 作之一m ®第2圖為本發明影像位移偵測操 乍之方法只細例的示意圖。本發明爭伤 含有下列㈣： _本發明㈣位移偵測操作包 :驟2〇〇:接收一符合pal規格的合成訊號； ν驟210 :根據目標圖框 判斷目椤圖拒P 、、鄰近圖框所攜帶的資訊，來 、，斷“圖框上-目標像素的位移狀況。 首先’影像位移摘測3|舍接此人L 號係符合PAL規格，』η:合成訊號。該合成訊 框(步驟·)。接著，圖框與至少-鄰近圖 谈者衫像位移偵測器便會由目鉀圖拖伽 ::)框所攜帶的資訊’來判斷—目標像素的位移:況(步 ^ .1 # ^ 11 m ^ ^ ^ I狍例的不意圖。纟包含有下列步 12 200908758 驟： 步驟300 .接收一符合PAL規格的合成訊號； 步驟3 10 :取得一目標像素之像素值； 步驟320 :從一鄰近圖框取得與該目標像素同相之對應像 素的像素值； 步驟330 ·若該對應像素的像素值與該目標像素的像素值 的差量小於一預定閥值；則接至步驟34〇，若否，則 接至步驟350; 步驟340 :判斷該目標像素為一靜止像素； 步驟3 50 :判斷該目標像素非一靜止像素。 於本實施例中，於接收到符合PAL規格的合成訊號後 (步驟300)，影像位移偵測器可以取得目標像素的像素值 ::梯^0) ’並且從—鄰近圖框取得與該目標像素同相之 對應像素的像素值（步驟32())。理論上， 有靜止（still)的性質，那麻知抑 心像素具 同.因奸鄰兩圖框之間的像素值應該相 及參考像素的像素值，若兩像素 =的像素值以 閥儐f斗^ 的差里小於一預定 閥值(步㈣〇)，那麼便可判斷預疋 質（步驟340)，戋否，π 3 象素八有靜止的性 驟350)。 則该目標像素並不為-靜止像素(步 在此請注意 非本發明的限制 像素來判斷目標 為判斷的依據； 叩做法僅為本發 。在眘ι β吧列，而 牙、應用上，本發明可以利 像素的位務壯更夕的 位移狀況，而無須僅 舉例來々 I * 曰&像素作 干〜木忒，本發明可 乃J以利用—個3*3像素 13 200908758 的影像區域（其中目標像素位於 砂1豕區域的中心』，以 藉由目標像素及/或其周圍像素的相關資訊來判斷目Μ 素的位移狀況，如此的相對應變化，亦屬本發明的範:。 =來說，請參閱第4圖’第4圖為本發明影像位移 债測刼作之另一方法實施例的示意圖。其包含有下列步 驟· 步驟彻：接收一符合pAL規格的合成訊號； 步驟410 :以目標像素為中心 像區域； 具有3*3像素的影 步驟420 :對於該影像區域 m Q ± 1豕果仗一鄰近圖框取 β玄目軚像素同相之對應像素的像素值； 步驟430 ··計算該影像區此 ^ 傢素與其於步驟420中 所取件之對應像素的像素值差量，並且 複數個差量進行加權平 异出來之 值； ；埯异並產生—加權平均 步驟440:若該加權平 450，若$目，社預定間値，則接至步驟 5〇右否，則接至步驟460; =驟450··判斷該目標像素為一靜止像素； 步驟彻：判斷該目標像素非—靜止像素。 於本實施例中，於接收到符合ρ (步驟400)，本發明 袼的s成訊號後 區域，其為-個判斷目^像^谓測器會先取得-個參考 所習知，若該目標像素係為兄^準。如業界 的周圍區域大致上也應具有=該目標像素 耵注貝因此，於本實施 14 200908758 例中’本發明絲得對應於該目 例來說，本發明可利用— 素的一參考區域，舉 標像素係位於該影像區域的中心素：：像區域’其中該目 標像素的周圍像素（步驟 。& 餘八個像素為該目 所廣泛知悉，素之參考區二；為如:舰項技術者 之限制條件，亦即其他大 A —例，而非本發明 明。 ，考區域亦可用來實現本發 移狀二參考秦域的位 框=Γ像素皆可由其相鄰圖框(可能是前-圖 施二到—對應像素與之同相；因此，於本實 J Ύ對於參考影像區域中的夂... 写皆舍*豕域中的母—像素，影像位移债測 ::由：-圖框或後—圖框取得—同相的對應像素(步 且計算對應像素與影像區域中像素的像素值差 =體來說’於參考區域中…9個像素中5 個像素（即目標像素本身及位於四個角落的4個像素）可 以於前-圖框（於另—實施例中，後—圖框）中找到盘其 同相之對應像素，而另外4個像素（即目標像素之上下左 右4個像素）可以於後一圖框（於另一實施例中，前一圖 框）中找到與其同相之對應像素。 對於目標像素而言，3*3的參考區域中每一個像素的 重要性應該有所不同，舉例來說，既然:是判斷目標像素本 身的4移狀况’與目標像素本身相關的的資訊理應具有最 问的重要性（於本實施例中，此資訊係指該目標像素與其對 15 200908758 200908758 而位 於 應像素的像素值差量)，上下左右四個 於四個角落的四個像素則具有較低的參 厂人之’市 本實施例中，本發明係對於之前 因此，於 差量進行加權平均運算，以取 °异的複數個像素值 Λ 4卜往焯立4城Τ 亏加權平均值（步驟4 3 0) 〇 在此❸主思，加權平均的概念係用來 ) 的參考性，此做法並非本發明的限制。舉㈣說m 明之另一實施例中，亦可將參考 ；本毛 設定為1，此即代表將每—個禮:每—像素的權重均 均運算’以得出-差量的平均值·，或者，：;直::： ，來之各個差量之絕對值相加，亦即計算絕對差量:: :贿。fabsolutedifferences，SAD)，作為與間值比:之 參數。如此的相對應變化，亦屬本發明的範轉。 ▲最後，本發明係將該加權平均值與一預定闕值進行比 較(步驟440)。如前所述，若目標像素具有靜止的特性， 那麼加權平均值理論上應該不會很大。因&，若加權平均 值小於該預定閥值，那麼_斷目標像素係為—靜止像素 (步驟450)，反之，則判斷目標像素並非為一靜止像素（步 驟 460)。 在此請注意，於前面的揭露中，由相鄰圖框取得相位 相同之對應像素的操作，對於此領域具有通常知識者並不 為難；如業界所習知，當影像解碼器接收到合成訊號時， 便會於合成訊號的特定位置，取得一用於調變彩度訊號之 栽波訊號（color burst)，該载波訊號係用來提供相位資訊， 以用於解調變操作；因此，影像解碼器便可以由該載波訊 16 200908758 號取得其後影像資料的相位資訊。換言之，本發明便可以 的機制，取得每—圖框的相位，也藉此而得知目 仏圖框與鄰近圖框的相位對應關係。舉例來說，藉 的機制，本發明可以得知目伊 述 框同相，而哪些像素係與後一圖框同相。 _ 在此請參閱第5圖，第5圖為本發明 5〇〇之-實施例的示意圖。如第 杪谓心 5» ,ππ θ . _ 7 第5圖所不，影像位移偵測 益500具有二個像素值差量計算模組51〇、53〇、别，一 加權運算模組570,以及一判斷模組58〇。在此請注音， 影像位㈣測器500的操作係對應第4圖所示的步驟，而 各疋件的功能與操作將於以下的揭露中詳述。 如前所述，於本實施例中，本發明係利用- 3*3的參 考影像區域來判斷目標像素的位移狀況。於第5圖中，像 素值差量計算模組51〇、53()、55()係分別用來處理參考影 像區域的各—掃描線。請注意，雖然於第5圖中並未繪出， 但是熟習此項技術的人應可理解，像素值差量計算模組 10 530 550通常係輕接於位於其前方之—緩衝器，並 且自該緩衝器中取出計算所需之像素值資訊。 _如第5圖所示，像素值差量計算模組510係分別接收 前一圖框F1 ’目標圖框F2，與後一圖框f3的前一掃描線 line(N-l)，像素值差量計算模組53()係分別接收前一圖框 F1 ’目標圖框F2 ’與後一圖框F3的目標掃描線line(N); 而像素值差夏計算模組55〇係分別接收前一圖框F1，目標 圖框F2，與後一圖框F3的後一掃描線line(N+1)。 17 200908758 於像素值差量計算模組510之_，加法器511係用來 計算前一圖框F1之前一掃描線πηε(Ν_υ上一像素與目標 圖框F2之前一掃描線丨inewq)上一對應像素之間的像素 值差值，並且由其後的絕對值計算模組513進行絕對值運 算’以計算出兩像素之間的差量。 此外，延遲單元515'517係分別用來延遲一像素； 因此’輸入至多工3 521、522、523的數值係為前—圖框 F1之前一掃描線上三個像素（即為3*3參考區域 中的上面的三個像素）與目標圖框F2之前一掃描線 line(N-l)上三對應像素的像素值差量。 田、、、 目同地，加法器512係用來計算目標圖框F2之前一 掃描線line(N-l)上一像素與後一圖框F3之前一掃描線 心㈤）上-對應像素之間的像素值差值，並且由其^的 絕對值計算模組514進行絕對值運算，以計算出兩像素之 間的差量。 'Composite signal^ Y+Usin(wt)±Vcos(wt) 200908758 It can be seen from the above equation (2) that the PAL image system defines that the carriers of the upper and lower horizontal scanning lines are 90 degrees out of phase, in other words, the nth scanning line and the ηth The carrier of the +2 scan lines differs by only 8 degrees, and since each frame has 62 5 scan lines in the pal system, the same scan lines of the front and rear frames are only 90 degrees apart, that is, the nth The line of the fourth line of the frame is only 18 degrees away from the carrier of the line of the claws of the n+2th frame. In other words, in the PAL system, if you want to use the two frames with the same phase chroma carrier for image motion detection, you must use the information of the nth frame and the n+4 frame to complete. However, such an approach is less difficult in the NTSC system. The TSC system only needs to temporarily store two frames, such as the nth frame and the state=frame' plus the n+2 m frame. The composite signal of the scan lines of several scan lines can be used to detect the displacement of the image. Therefore, for the buffer (4) (4) = the capacity requirement is not too large; however, in the 纟PAL system, if the image is to be processed by the phase:: method For displacement detection, you need to temporarily store four frames, j such as the nth frame to the n+3th image ^, take the frame, and add the number of lines in the n+4 frame. Synthesize the signal data, and start to serve. Obviously, the capacity requirements of the PAL·糸··中中’ buffer are not economical. The SC is twice as large as the [Summary of the Invention] Therefore, the main object of the present invention is to provide an image displacement topping method, which can utilize less image in the image image displacement of the pAL image. The frame reaches the problem of too much shadow. ▲ The capacity of the buffer in the technology is required to be 200908758. According to an embodiment of the invention, an image displacement detecting method is disclosed, which is applied to a PAL image system, and the image displacement detecting method includes: receiving a composite signal ( Composite signal), the composite signal carrying a target frame and at least a first reference frame adjacent to the target frame; and the first information and the first reference carried according to the target frame The frame carries a first corresponding information to determine a displacement condition of the target pixel of the target frame; wherein the first asset has the same chroma carrier phase as the first corresponding information. According to an embodiment of the present invention, an image displacement detecting method is applied to a PAL image system, and the image displacement detecting method includes: receiving a composite signal, the synthesizing The signal carries a target frame (frame) and a first reference frame and a second reference frame adjacent to the target frame; for one target target of the target frame, one of the corresponding target pixels is obtained. An image area, the image area: 2 has a plurality of pixels; for each pixel of the plurality of pixels, a corresponding pixel having the same chroma phase is obtained from the first reference frame or the second reference frame; Each of the plurality of pixels—(4) a difference in pixel value of the corresponding pixel f to obtain a plurality of pixel value differences; performing a weighted average operation on the pixel value difference to obtain a weighted average; and according to the weighted average , determining the displacement condition of the target pixel. According to one embodiment of the present invention, an image displacement debt detector is disclosed: the image displacement detection method includes: receiving a composite signal, the synthesis The signal carries a target frame (frame) and a reference frame adjacent to the target frame _ 9 200908758 and a second reference frame; for a Ί, a know pixel, obtain a plurality of corresponding δ hai target pixels The reference image is said to be the first information carried by the first reference frame and the first reference frame carried by the first reference frame, and one of the first frames according to the target frame. The second information material = the second corresponding information carried in the test frame to determine the reference displacement state of the plurality of materials; (4) the reference displacement state of the plurality of pixels to determine the displacement state of the target pixel; wherein the The information has the same chroma carrier phase as the first correspondence information, and the second information has the same chroma carrier phase as the second corresponding information. / The image detection method of the invention performs image displacement detection in the PAL image system, and does not need to temporarily store four frames, but only needs to be πI瞀孖-to two frames to achieve displacement debt measurement (4), Because of this, this #明 has a smaller capacity requirement for the buffer, which can save more cost. [Embodiment] Referring to Fig. 1, Fig. 1 is a diagram showing pixel values of the same scanning line μ of different frames (N-2 to N+2) in the PAL image system. In the figure i, it is assumed that each scanning line is sampled at 45, 丨3 5, 225, 3 15 degrees of the carrier. Moreover, for the sake of simplicity, each scanning line shows only four pixels, and The pixel value is also indicated next to each pixel, especially the phase relationship of its carrier. Note here that in Figure 1, the pixels of each frame N_2 to frame n+2 correspond to each other. In other words, the following pixels: (pi, p5, P9, P13, P17); (P2) P6, P10, P14, P18); (P3, P7, P11, P15, P19); (P4, P8, P12, P16, P20), which are the same positions on the range of 200908758 to n+2, respectively. In addition, the chroma value of the pixel value (for example, the pixel value Yi+(V1+U1) of the pixel pi) represents the luminance value of the money, and the towel V and U represent the pixel by the first image. It can be seen that since the carrier phases of two adjacent frames are different by 90, therefore, only the frame & disk frame is completely the same in the carrier phase - however, there is still a corresponding relationship between the two adjacent frames. In the frame Μ. The frame of the chorus frame is not exactly the same as the whole scan line: but the pixel P5 and the pixel p9, the pixel p7 and the pixel (1) are in the same H plane, with the frame N+1 and the frame N are exemplified. Although the entire scanning line is not completely in phase with each other, the pixel P1 〇 and the pixel P14, the pixel P12 and the pixel p16 are in phase. In fact, among the scan lines of the frame N, half of the pixels can find the corresponding pixels in the same frame N-1, and the other half of the pixels can be in the subsequent frame N+1. Find the corresponding pixel in the same phase. For example, taking frame N as an example, half of the pixels p9 and pu are respectively in phase with the corresponding pixels P5 and P7 of the previous frame, and the other half of the pixels ρι〇 and P12 are the same as the next frame N+1. The corresponding pixels pu and pi6 are in phase. In the frame N, the half pixels which are in phase with the corresponding pixels of the previous frame and the other half pixels which are in phase with the corresponding pixels of the next frame N+1 are alternately arranged. As described above, if the image displacement detection operation is to obtain the best effect, the corresponding pixels with the same carrier phase in different frames must be used. In the prior art, such properties depend on each reference. Corresponding pixel load 11 200908758 Binocular position::::Two: box (such as the above frame N and frame N+4), frame (4) pre-fetching f, direct «box N adjacent frame For example, the solid 4, τ, ^ W state burns the frame Ν-1 and the frame N+1, the confrontation, the (four) (four) (four) ^ take (four) wave phase the same figure: 5N: for the target of the frame U Pixels, the present invention is only required to be: the target: phase: frame - or one, the displacement can be achieved with two = two. As can be seen from the foregoing, the present invention can replace the conventional technology to make it clear: frame and frame Ν +4 way of doing. Therefore, in the case of the second-shift detection, only two temporary storages need to be temporarily stored, thereby saving half of the data saved by the slow two-ton, that is, the construction cost of the buffer state. Please refer to Fig. 2, 之一m m ^ 2 is a schematic diagram of a detailed example of the method of image displacement detection operation of the present invention. The present invention has the following (4): _ The present invention (4) Displacement detection operation package: Step 2: Receive a composite signal conforming to the pal specification; νStep 210: Determine the target map rejection P according to the target frame, and the adjacent map The information carried in the box, come, and break the "frame-to-target pixel displacement status. First of all" image displacement measurement 3 | rounded up this person L number is in line with PAL specifications, η: composite signal. The composite frame (Step ·). Next, the frame and at least the neighboring viewer's shirt displacement detector will be judged by the information carried by the target::) the displacement of the target pixel: ^ .1 # ^ 11 m ^ ^ ^ I do not intend to use the following example. Step 12: 200908758 Step: Step 300: Receive a composite signal conforming to the PAL specification; Step 3 10: Obtain a pixel value of a target pixel; Step 320: Obtain a pixel value of a corresponding pixel that is in phase with the target pixel from a neighboring frame. Step 330: If the difference between the pixel value of the corresponding pixel and the pixel value of the target pixel is less than a predetermined threshold, then Step 34: If no, proceed to step 350; Step 340: determine the The target pixel is a still pixel; Step 3 50: determining that the target pixel is not a still pixel. In this embodiment, after receiving the synthesized signal conforming to the PAL standard (step 300), the image displacement detector can obtain the target pixel. Pixel value:: ladder ^0) 'and obtain the pixel value of the corresponding pixel in phase with the target pixel from the adjacent frame (step 32 ()). Theoretically, there is still property, then The pixel pixel has the same pixel value as the reference pixel. If the pixel value of the two pixels = the difference between the valve and the valve is less than a predetermined threshold (step (4)〇 ), then the pre-quality can be judged (step 340), 戋 no, π 3 pixel eight has a static polarity 350). Then the target pixel is not a static pixel (step here please note that the invention is not Limiting the pixel to judge the target as the basis for judgment; 叩The practice is only the present hair. In the case of caution, the invention can improve the displacement of the pixel, and it is not necessary to just give an example. 々I * 曰 & pixels for dry ~ hibiscus, the invention can J uses a 3*3 pixel 13 200908758 image area (where the target pixel is located at the center of the sand 1豕 area) to determine the displacement of the target by the relevant information of the target pixel and/or its surrounding pixels. Such a corresponding change is also a mode of the present invention: =, for example, please refer to FIG. 4'. FIG. 4 is a schematic diagram of another method embodiment of the image displacement debt measurement method of the present invention. Step · Step: Receive a composite signal conforming to the pAL specification; Step 410: center the image area with the target pixel; shadow with 3*3 pixels Step 420: For the image area m Q ± 1 Taking the pixel value of the corresponding pixel of the same phase of the pixel; step 430 · calculating the pixel value difference between the pixel and the corresponding pixel of the component taken in step 420, and weighting the plurality of differences The value of the difference is different; the difference is generated - weighted average step 440: if the weighting is 450, if the order is $, if the order is 値, then it is connected to step 5 〇 right, then to step 460; ··Judge the target pixel as Still-pixel; Toru steps: determining whether the target pixel is a non - stationary pixels. In this embodiment, after receiving the ρ (step 400), the s of the present invention is a post-signal region, which is a judgment target, and the pre-detector first obtains a reference, if The target pixel is a brother. For example, the surrounding area of the industry should also have = the target pixel. Therefore, in the example of the present invention, in the case of the present invention, the present invention can be utilized as a reference area of the present invention. The target pixel is located in the center of the image area: the image area 'where the surrounding pixels of the target pixel (steps & the remaining eight pixels are widely known for the purpose, the reference area of the prime; for example: the ship The limitations of the technicians, that is, other large A-examples, rather than the present invention, can also be used to implement the bit-frame of the second-order reference domain of the present invention: the pixels can be adjacent to the frame (possibly It is the front-figure two to - the corresponding pixel is in phase with it; therefore, in this real J Ύ for the reference image area... Write all the * mother-pixel in the field, image displacement debt test:: by :- Frame or Back - Frame Get - the corresponding pixel of the same phase (step and calculate the pixel value difference between the corresponding pixel and the pixel in the image area = body] in the reference area... 5 pixels out of 9 pixels (ie target The pixel itself and 4 pixels in four corners The corresponding pixels of the same phase can be found in the front-frame (in the other embodiment, the back-frame), and the other 4 pixels (ie, the upper and lower 4 pixels of the target pixel) can be in the next frame. (In another embodiment, the previous frame) finds the corresponding pixel in phase with it. For the target pixel, the importance of each pixel in the 3*3 reference area should be different, for example, since : is to determine the 4 shift status of the target pixel itself. The information related to the target pixel itself should have the most important importance (in the present embodiment, this information refers to the target pixel and its pair 15 200908758 200908758 and is located in the pixel The pixel value difference), four pixels above and below the four corners of the four corners have a lower quotient of the factory. In the present embodiment, the present invention performs a weighted averaging operation on the difference for the previous Take a multiplicity of pixel values of ° 4 焯 4 焯 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 The invention In another embodiment, the reference may also be referred to; the local hair is set to 1, which means that each of the rituals: the weight of each pixel is averaged to obtain the average of the difference. The value ·, or,:; straight::: , the absolute value of each difference is added, that is, the absolute difference is calculated:: : bribe. fabsolutedifferences, SAD), as the ratio to the ratio: the parameter. Corresponding changes are also a paradigm of the present invention. ▲ Finally, the present invention compares the weighted average with a predetermined threshold (step 440). As previously mentioned, if the target pixel has a stationary characteristic, then the weighting The average value should not be large in theory. Because &, if the weighted average is less than the predetermined threshold, then the target pixel is - still pixel (step 450), otherwise, the target pixel is not a still pixel. (Step 460). Please note that in the foregoing disclosure, the operation of obtaining corresponding pixels of the same phase from adjacent frames is not difficult for those having ordinary knowledge in the field; as is known in the art, when the image decoder receives the synthesized signal At the specific position of the composite signal, a color burst for adjusting the color signal is obtained, and the carrier signal is used to provide phase information for demodulating the variable operation; therefore, the image The decoder can obtain the phase information of the subsequent image data from the carrier signal 16 200908758. In other words, the present invention makes it possible to obtain the phase of each frame, and thereby know the phase correspondence between the target frame and the adjacent frame. For example, by means of the borrowing mechanism, the present invention can know that the frame of the object is in phase, and which pixels are in phase with the latter frame. _ Referring to Figure 5, Figure 5 is a schematic view of an embodiment of the present invention. For example, the third point is 5», ππ θ. _ 7 Figure 5 does not, the image displacement detection benefit 500 has two pixel value difference calculation modules 51〇, 53〇, another, a weighting operation module 570, And a judging module 58〇. Note here, the operation of the image bit (four) detector 500 corresponds to the steps shown in Figure 4, and the function and operation of each component will be detailed in the following disclosure. As described above, in the present embodiment, the present invention judges the displacement state of the target pixel by using the reference image area of -3*3. In Fig. 5, the pixel value difference calculation modules 51A, 53(), and 55() are used to process the respective scanning lines of the reference image area. Please note that although not depicted in FIG. 5, those skilled in the art will appreciate that the pixel value difference calculation module 10 530 550 is typically lightly coupled to the buffer located in front of it, and The buffer stores the pixel value information required for the calculation. As shown in FIG. 5, the pixel value difference calculation module 510 receives the previous frame F1 'target frame F2 and the previous scan line line (N1) of the subsequent frame f3, and the pixel value difference The calculation module 53() receives the target scan line line(N) of the previous frame F1 'target frame F2' and the subsequent frame F3, respectively; and the pixel value difference summer calculation module 55 receives the previous one. Frame F1, target frame F2, and the next scan line line (N+1) of the next frame F3. 17 200908758 In the pixel value difference calculation module 510, the adder 511 is used to calculate a scan line πηε (the previous pixel and the previous scan line 丨inewq before the target frame F2) before the previous frame F1. The pixel value difference between the pixels is corresponding, and the absolute value calculation module 513 performs the absolute value operation ' to calculate the difference between the two pixels. In addition, the delay unit 515'517 is used to delay one pixel respectively; therefore, the value of the input to the multiplex 3 521, 522, 523 is three pixels on the scan line before the front frame F1 (ie, the 3*3 reference area) The upper three pixels in the middle) are different from the pixel values of the three corresponding pixels on the scan line line (N1) before the target frame F2. Field, and, similarly, the adder 512 is used to calculate between a pixel on the scan line line (Nl) before the target frame F2 and on the scan line (5) before the next frame F3. The pixel value difference is calculated by the absolute value calculation module 514 of the absolute value calculation to calculate the difference between the two pixels. '
/. I 延遲单元516、518係分別用來延遲 因此：輸入至多工請、522、⑵的數值係為目標圖框 F2之别—掃描、線ηη·υ上三個像素（即為3*3參考區域 中的上面的三個像素）與後一圖才匡F3 <前一掃描線 line(N-l)上三對應像素的像素值差量。 ^此了來’輸入至多工器521、522、切兩輸入端的 訊號》別為目標圖框與兩相鄰圖框（包含前一圖框以及後 -圖框）上兩個對應像素的差量；如前所述，在PM規格 中’任一像素皆能從相鄰兩圖框中的一圖框中找到同相位 18 200908758 的對應像素；因此，選擇訊號PAL_mod係用來控制多工器 521 522、523的輸出，以選擇輸出兩個同相之像素的差 量。具體來說’當PAL—mod = I時，多工器52 1、522、523 會輸出位於輸入端，，】，，之訊號；當PAL—m〇d = 〇時，多工 器52〗、522、523會輸出位於輸入端，，〇，，之訊號。 在此請注意，於本實施例中，像素值差量計算模組 530、550與像素值差量計算模組51〇具有相同的功能與運 作，故不另贅述於此。 當3*3參考區域的每一像素與其同相的對應像素之像 素值差量皆由多工器521〜523、541〜543、561〜563輸出之 後，加權平均模組570便會對這些像素值差量進行加權平 均運算。如第5圖所示，於本實施例中，目標像素所對應 的權重為4而其上下左右的相鄰像素的權重為2,其餘 位於角落的四個像素所對應的權重則4 1。在經過乘法器 571〜575的加權運算（乘法運算）以後，加法器會將乘 法器57L的運算結果相加，接著，平均器m會將加 ^器576所計算出來的總和加以平均（於本實施例中，平均 器577係將加法器576所計算出來的總和數值除以1 6)。 而熟習此項技術的人應可理解，此處平均器m僅為一選 擇性（optional )之元件，而非必要。 最後’判斷模組580便會將平均器577所輸出的加權 #值’與-預㈣值進行比較。如前所述，若該加權平 W小於該預定閥值’則代表目標像素為-靜止像素；反 之則代表目標像素並非為-靜止像素。 、 19 200908758 在此晴 >主意，於前 可利用硬體電路實作之 ' :實施例中’雖然所有的操作皆 -實施例，而非本發明^而’這樣的做法僅為本發明之 作皆可利用—處理器執二制。在實際應用上，前述的操 化，亦屬本發明的範/❺體實施之，如此的相對應變 此外’於前述的眘 應的參考區域來決定目^ ’本發明利用目標像素所對 做法亦僅為本發明之的位移狀況。然而’這樣的 來說，本發明亦可先以:::考:::::的限… 與其周圍像素的位移狀況，接^ =的方法判斷目標像素 素的位移狀況作為參考再將目標像素與其周圍像 移狀況。 進而辑出該目標像素的最終位 更明確地說，本發明八 素為中心，分別計算出目心目標像素與其周圍像 平均值，以及周圍像素所分 =參考£域的加權 值，之後再依據這此加權平约值考區域的加權平均 狀況（譬如將這些加權;Γ值再進L來判斷目標像素的位移 這些加權平均值）。 再進仃—次加權運算，來判斷 揭路至此’此領域具有诵當 之，故苴相斟廊举罢* 吊知識者應可理解並實施 “對應裝置並不另贅述於此。舉例來說 籌多個前述的像素值差量計算模組 :、便、 以計算出多個像素之參考區域的 便： 另逵罢β /«, 卞句數。接者，只須 器)，並可將模組(譬如可另建置-個低通遽波 刚述的夕個加權平均數進行運算，以得出-個 20 200908758 判斷的依據；最後，判斷模組便可以將此判斷依據與—預 定閥值進行比較，便可進行位移狀況的判斷。 相較於習知技術，本發明之影像偵測方法於pal影像 系統中進行影像位移偵測時，無須暫存四個圖框，而只須 要暫存一至兩個圖框即可達成位移偵測的目的，因此，本 發明對於緩衝器（buffer)的容量需求較小，可以節省更多的 成本。 μ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。 【圖式簡單說明】 第i圖為PAL影像系統中不同圖框之同—掃描線的像 素值示意圖。 第2圖為本發明影像位移偵測操作之一方法實施例的 不意圖。 第3圖為本發明於像位移偵測操作之另一方法實施例 的不意圖。 第4圖為本發明影像位移偵測操作之另一方法實施例 的不意圖。 第5圖為本發明影像位移偵測器之一實施例的示意 圖0 圖式編號 500 位移判斷模組 510 、 530 、 550 像素值差量計算模組 21 200908758 511、512、531、532、551、552、566 加法器 513、514、533、53 4、553、554 絕對值運算模組 515〜518、535〜538、555〜558 延遲單元 521〜523 、 541〜543 、 561〜563 多工器 570 加權平均模組 571〜575 乘法器 577 平均器 576 加法器 580 判斷模組 22/ I I delay units 516, 518 are used to delay, therefore: the input to the multiplex, 522, (2) the value is the target frame F2 - scan, line ηη · υ three pixels (that is 3 * 3 The upper three pixels in the reference area) and the latter figure are the difference between the pixel values of the three corresponding pixels on the previous scan line line (N1). ^This is the input signal to the multiplexer 521, 522, and the two input terminals. The difference between the two corresponding pixels on the target frame and the two adjacent frames (including the previous frame and the back frame). As mentioned above, in the PM specification, 'any pixel can find the corresponding pixel of the same phase 18 200908758 from a frame in the adjacent two frames; therefore, the selection signal PAL_mod is used to control the multiplexer 521 The output of 522, 523 is selected to output the difference between the two pixels in phase. Specifically, when PAL-mod = I, the multiplexer 52 1 , 522 , 523 will output the signal at the input end, ,,,; when PAL-m〇d = ,, the multiplexer 52〗 522, 523 will output the signal at the input, 〇,,. Please note that in the present embodiment, the pixel value difference calculation modules 530 and 550 have the same functions and operations as the pixel value difference calculation module 51A, and therefore will not be further described herein. When the pixel value difference between each pixel of the 3*3 reference area and the corresponding pixel of the same phase is output by the multiplexers 521 to 523, 541 to 543, and 561 to 563, the weighted average module 570 will have these pixel values. The difference is weighted averaged. As shown in Fig. 5, in the present embodiment, the weight corresponding to the target pixel is 4, and the weight of the adjacent pixels on the top, bottom, left, and right is 2, and the weight of the remaining 4 pixels in the corner is 4 1 . After the weighting operation (multiplication operation) by the multipliers 571 to 575, the adder adds the operation results of the multiplier 57L, and then the averager m averages the sum calculated by the adder 576 (in this case). In the embodiment, the averager 577 divides the sum value calculated by the adder 576 by 16). Those skilled in the art should understand that the averager m is only an optional component, not a necessity. Finally, the decision module 580 compares the weighted #value' output by the averager 577 with the -pre-four value. As previously mentioned, if the weighted flat W is less than the predetermined threshold ', the target pixel is a -still pixel; and the target pixel is not a - still pixel. 19 200908758 Here, the idea is that the hardware circuit can be used in the prior art: in the embodiment, although all the operations are the embodiments, not the present invention, the practice is only the present invention. All can be used - processor implementation. In practical applications, the aforementioned operation is also implemented by the invention. The relative strain is also determined by the aforementioned reference area. It is only the displacement condition of the present invention. However, in this case, the present invention can also firstly use the ::: test::::: limit... and the displacement of the surrounding pixels, the method of ^ = determines the displacement of the target pixel as a reference and then the target pixel It is like moving around. Further, the final bit of the target pixel is set. More specifically, the octet of the present invention is centered, and the average value of the centroid target pixel and its surrounding image is calculated, and the weight value of the surrounding pixel is divided into the reference field, and then The weighted average of the weighted flat values is (for example, weighting these; the threshold is further incremented by L to determine the weighted average of the displacements of the target pixels). Then enter the 仃-time-weighted operation to judge the road to this. This area has a slap in the face, so the 苴 斟 * * * 吊 吊 知识 知识 知识 知识 知识 知识 知识 知识 知识 知识 知识 对应 对应 对应 对应 对应 对应 对应 对应 对应 对应 对应 对应 对应 对应 对应 对应A plurality of the aforementioned pixel value difference calculation modules are: for example, to calculate a reference area of a plurality of pixels: another ββ /«, a number of haikus, a receiver, only a device, and The module (for example, another built-in low-pass chopper just described the weighted average of the evening to calculate the basis for the judgment of 20 200908758; finally, the judgment module can judge the basis and the reservation When the threshold is compared, the displacement condition can be judged. Compared with the prior art, the image detection method of the present invention does not need to temporarily store four frames when performing image displacement detection in the pal image system, but only needs to be stored. The purpose of the displacement detection can be achieved by temporarily storing one or two frames. Therefore, the present invention has less capacity requirements for buffers and can save more cost. Although the present invention is described by way of examples, It is not intended to limit the scope of the invention The industry can perform various modifications or changes without departing from the gist of the present invention. [Simple description of the drawing] The i-th picture is the same as the pixel value of the different lines in the PAL image system. The invention is not intended to be an embodiment of the image displacement detecting operation. Fig. 3 is a schematic diagram of another embodiment of the image displacement detecting operation of the present invention. The fifth embodiment is a schematic diagram of an embodiment of the image displacement detector of the present invention. The figure number 500 displacement determination module 510, 530, 550 pixel value difference calculation module 21 200908758 511, 512, 531, 532, 551, 552, 566 adders 513, 514, 533, 53 4, 553, 554 absolute value calculation modules 515 to 518, 535 to 538, 555 to 558 delay units 521 to 523, 541 to 543 561~563 multiplexer 570 weighted average module 571~575 multiplier 577 averager 576 adder 580 judgment module 22